Method and device for inspection of liquid articles

[0001] The present application claims priority of Chinese patent application Serial No. 200710180653.2, filed Oct. 5, 2007, the content of which is hereby incorporated by reference in its entirety.

[0002] The present invention is related to the field of radiation inspection technique, and more particularly to a method and device for quick security-inspection of liquid articles with dual-energy CT.

[0003] Since 9•11 in U.S., security-inspection of aviation is becoming more and more emphasized. Besides conventional security-inspection of packages, security-inspection of the liquid articles carried by passengers is added. Accordingly, it is in dire need of means and methods for quick security-inspection of the liquid articles.

[0004] Nowadays, there are four types of methods used in security-inspection of liquid articles, including chemical method, electromagnetic method, neutron detection method and radiation detection method as follows:

[0005] 1) The chemical method can be subdivided into odor identification method, ion scanning explosive detection method and substance analysis method. The odor identification in practical applications often fails to implement detection because of sealed and packaged conditions of liquid articles. The ion scanning explosive detection method is known for its high sensitivity, but with high false alarm rate, it suffers from the affection of background environment. The substance analysis is of high precision and high accuracy, but this method needs a certain period of time to analyze the sample, which cannot meet the demands of on-site quick detection.

[0006] 2) The electromagnetic method works in an active measurement manner. It distinguishes liquid articles from each other according to their dielectric constants in an electromagnetic field. The electromagnetic method is easily subjected to severe affection of metal packages or other thick material packages. As a result, the use of electromagnetic method is limited in case of complex package materials.

[0007] 3) The use of the neutron detection method will leave residual radiation remaining in the detected liquid due to the effect of “neutron activation”. Furthermore, the radiation shielding is complicate due to neutrons' strong penetrability, and the apparatus has to take a large area, so the method is not suitable for application in the security-inspection systems of civil aviation.

[0008] 4) Currently, most of the security-inspection apparatuses for civil aviation are radiation apparatuses. In these apparatuses, the 2D X-ray imaging technology and three-dimensional CT technology are mostly adopted. These technologies, which can obtain structure information of articles, fail to determine whether there are drugs concealed in the liquid articles. The reason is that the overall structure (for example, the number of layers) of the liquid articles will not change a lot with drugs being concealed therein, only the components will change.

[0009] To sum up, for the quick detection of the liquid articles, the chemical method, the electromagnetic method and the neutron detection method are not suitable. By using the 2D X-ray imaging technology and the three-dimensional CT technology, images including the structure information of the article are acquired, but these images cannot work as sufficient evidence for whether there are drugs concealed in the liquid articles.

[0010] In order to overcome the disadvantages in the existing technologies, an object of the invention is to provide a method as well as a device for security-inspection of a liquid article using radiations, which can conduct a quick detection and get quantitative information of the liquid article to be inspected, without destroying the outer packing.

[0011] In the first aspect of the invention, the invention provides a method for security-inspection of a liquid article with dual-energy CT, comprising the steps of: acquiring dual-energy projection data by dual-energy CT scanning on the liquid article to be inspected; performing CT reconstruction on the projection data to obtain a CT image which indicates physical attributes of the inspected liquid article; extracting the physical attributes of the inspected liquid article based on the CT image; and determining whether the inspected liquid article is suspicious according to the physical attributes and reference physical attributes of the inspected liquid article.

[0012] According to an embodiment of the invention, the physical attributes include the density and atomic number of the liquid article.

[0013] According to an embodiment of the invention, the dual-energy CT scanning works in a plane CT scanning manner.

[0014] According to an embodiment of the invention, the dual-energy CT scanning works in a normal spiral CT scanning manner.

[0015] According to an embodiment of the invention, the dual-energy CT scanning works in a high pitch spiral CT scanning manner.

[0016] According to an embodiment of the invention, a set of scan positions are preset prior to the plane CT scanning.

[0017] According to an embodiment of the invention, a DR scanning is performed to get a transmission image of the inspected article and the CT scan position is determined based on the transmission image, prior to the plane CT scanning.

[0018] According to an embodiment of the invention, after the transmission image has been gotten, the operator specifies at least one row of the transmission image via an input device as the CT scan position.

[0019] According to an embodiment of the invention, after the transmission image has been gotten, at least one row of the transmission image is automatically specified by the image processing technique as the CT scan position.

[0020] According to an embodiment of the invention, the step of getting the transmission image comprises emitting high-energy radiation beams and low-energy radiation beams which transmit the inspected article to form a high-energy transmission image and a low-energy transmission image; integrating the high-energy transmission image and the low-energy transmission image to form the transmission image.

[0021] According to an embodiment of the invention, the step of forming the transmission image comprises emitting high-energy radiation beams and low-energy radiation beams which transmit the inspected article to form a high-energy transmission image and a low-energy transmission image; selecting either the high-energy transmission image or the low-energy transmission image as the transmission image.

[0022] According to an embodiment of the invention, the step of performing CT reconstruction on the projection data to obtain a CT image which indicates physical attributes of the inspected liquid article comprises the steps of: generating projection data of two basis material coefficients based on the high-energy and low-energy projection data; performing reconstruction on the projection data of the two basis material coefficients to obtain a CT image which indicates the two basis material coefficients corresponding to the inspected liquid article; and generating a CT image indicating physical attributes of the inspected liquid article based on the CT image indicating the basis material coefficients.

[0023] According to an embodiment of the invention, the step of extracting the physical attributes of the inspected liquid article based on the CT image comprises the steps of: extracting pixels corresponding to the liquid article from the CT image; calculating the average density and atomic number of the pixels corresponding to the liquid article as the density and atomic number of the inspected liquid article.

[0024] According to an embodiment of the invention, the step of determining whether the inspected liquid article is suspicious according to the physical attributes and reference physical attributes of the inspected liquid article comprises the steps of: calculating the difference between the density and atomic number and reference density and atomic number; and determining there are drugs concealed in the inspected liquid article if the difference is larger than a predetermined threshold.

[0025] According to an embodiment of the invention, after dual-energy CT scanning at each of the positions, the CT images of the inspected liquid article are rotated to be aligned with the image formed after the first dual-energy CT scanning.

[0026] According to an embodiment of the invention, after dual-energy CT scanning at each of the positions, the inspected liquid article is rotated to the position before scanning.

[0027] According to an embodiment of the invention, several liquid articles are disposed in a barrel which is divided into multiple subspaces.

[0028] According to an embodiment of the invention, the method further comprises steps of: automatically detecting the presence of the barrel with a predefined pattern; determining a certain mark in the CT image in the case of the presence of the barrel; and rotating the barrel to a predefined position based on the certain mark.

[0029] According to an embodiment of the invention, the method further comprises the step of displaying determining result of the inspected liquid article on a display screen.

[0030] According to an embodiment of the invention, the method further comprises the step of printing determining result of respective liquid articles.

[0031] According to an embodiment of the invention, the method further comprises the step of colorizing the CT images of respective liquid articles.

[0032] In another aspect of the invention, the invention provides a device for security-inspection of a liquid article with dual-energy CT, comprising: a radiation source for emitting radiation beams; detection and collection means for detecting and collecting radiation beams transmitting at least one liquid article to be inspected; a controller for controlling the radiation source and the detection and collection means to perform dual-energy CT scanning on the inspected liquid article so as to obtain projection data; means for performing reconstruction on the projection data to obtain a CT image which indicates at least one physical attribute of the inspected liquid article; and means for determining whether the inspected liquid article is suspicious based on the physical attribute and reference physical attributes of the inspected liquid article.

[0033] According to an embodiment of the invention, the dual-energy CT scanning is performed on a predetermined position.

[0034] According to an embodiment of the invention, the detection and collection means detects and collects radiation beams transmitting the at least one liquid article to be inspected so as to form a transmission image; wherein the device further comprises means for specifying at least one row of the transmission image; and the dual-energy CT scanning is performed on the specified row.

[0035] According to an embodiment of the invention, the physical attributes includes at least the density and atomic number of the inspected liquid article.

[0036] According to an embodiment of the invention, the radiation source emits high-energy radiation beams and low-energy radiation beams which transmit the inspected article to form a high-energy transmission image and a low-energy transmission image; and the device further comprises means for integrating the high-energy transmission image and the low-energy transmission image to form the transmission image.

[0037] According to an embodiment of the invention, the radiation source emits high-energy radiation beams and low-energy radiation beams which transmit the inspected article to form a high-energy transmission image and a low-energy transmission image; and the device further comprises means for selecting either the high-energy transmission image or the low-energy transmission image as the transmission image.

[0038] According to an embodiment of the invention, the means for specifying at least one row of the transmission image comprises means for selecting at least one row by the operator from the transmission image via an input device.

[0039] According to an embodiment of the invention, the means for specifying at least one row of the transmission image comprises means for detecting liquid layers in the transmission image by analyzing pixels of the transmission image; and means for specifying central rows of respective layers as the rows to be performed dual-energy CT scanning.

[0040] According to an embodiment of the invention, the means for performing reconstruction on the projection data to obtain a CT image which indicates physical attributes of the inspected liquid article comprises means for integrating a density image identified by the density of the inspected liquid article and an atomic number image identified by the atomic number of the inspected liquid article to form a CT image; and means for extracting pixels corresponding to the liquid article from the CT image; and means for calculating the average density and atomic number of the pixels corresponding to the liquid article as the density and atomic number of the inspected liquid article.

[0041] According to an embodiment of the invention, the means of determining whether the inspected liquid article is suspicious according to the physical attributes and reference physical attributes of the inspected liquid article comprises means of calculating the difference between the density and atomic number and reference density and atomic number; and means of determining there are drugs concealed in the inspected liquid article if the difference is larger than a predetermined threshold.

[0042] According to an embodiment of the invention, the device further comprises means for, after dual-energy CT scanning at each of the rows, rotating the CT images of the inspected liquid article to be aligned with the image formed after the first dual-energy CT scanning.

[0043] According to an embodiment of the invention, the device further comprises means for, after dual-energy CT scanning at each of the rows, rotating the inspected liquid article to the position before scanning.

[0044] According to an embodiment of the invention, the device further comprises a barrel which is divided into multiple subspaces for disposing a plurality of liquid articles respectively.

[0045] According to an embodiment of the invention, the device further comprises means for automatically detecting the presence of the barrel with a predefined pattern; means for determining a certain mark in the CT image in the case of the presence of the barrel; and means for rotating the barrel to a predefined position based on the certain mark.

[0046] According to an embodiment of the invention, the device further comprises display means for displaying determining result of the inspected liquid article.

[0047] According to an embodiment of the invention, the device further comprises means for printing determining result of respective liquid articles.

[0048] According to an embodiment of the invention, the device further comprises means for colorizing the CT images of respective liquid articles.

[0049] According to an embodiment of the invention, the device further comprises a carrier mechanism to carry the liquid articles to be inspected, wherein the surface of the carrier mechanism on which the inspected liquid articles is carried is divided into a plurality of regions the operator can identify.

[0050] In yet another aspect of the invention, the invention provides a device for security-inspection of a liquid article with dual-energy CT, comprising a radiation source for emitting radiation beams; detection and collection means for detecting and collecting radiation beams transmitting at least one liquid article to be inspected; a controller for controlling the radiation source and the detection and collection means to perform spiral CT scanning on the inspected liquid article so as to obtain a set of spiral CT images each of which indicates at least one physical attribute of the inspected liquid article; means for analyzing the set of spiral CT images to acquire a spiral CT image part of the liquid article; and means for determining whether the inspected liquid article is suspicious based on the physical attribute and reference physical attribute of the inspected liquid article.

[0051] According to an embodiment of the invention, the physical attributes includes at least the density and atomic number of the inspected liquid article.

[0052] With the method and device according to the invention, the transmission image is used as a guide for the dual-energy scanning, and thus the detection speed can be improved without lowing detection accuracy. Furthermore, it can be determined whether the liquid article has an interlayer by means of the transmission image.

[0053] Moreover, it can be determined whether there are drugs (for example, cocaine etc.) concealed in liquid articles (for example, alcohol) by comparing the measured density and atomic number with reference density and atomic number.

[0054] In addition, inspection operation is facilitated because the operator can specify any position to perform the dual-energy CT scanning.

[0055] Further, a divided barrel is used when a plurality of articles is to be inspected, and so it can be easily determined which one of the liquid articles is suspicious.

[0088] The preferred embodiments of the invention will now be described more fully hereinafter with reference to the accompanying drawings. In the drawings the same reference numerals are used for denoting the same or similar components that are shown in different figures. For clarity, the detailed description of the known function and structure incorporated herein will be omitted, which would otherwise weaken the subject of the invention.

[0089] FIG. 1 is a schematic diagram of the structure of an inspection device according to an embodiment of the invention.

[0090] As shown in FIG. 1, the inspection device according to the invention comprises a radiation source 10 for emitting dual-energy X-rays for inspection, e.g. a X-ray machine; a carrier mechanism 40 which carries the liquid article 20 to be inspected and can rotate around axis Z thereof and can ascend or descend to take the liquid article 20 into the inspection area, thereby the radiations emitted by the radiation source 10 can transmit through the inspected liquid article 20; a detection and collection means 30 being an integrated module of a detector and a data collector, which is used to detect the dual-energy radiations transmitted through the liquid article 20 to acquire analog signals, and convert the analog signals into digital signals, and hence output the scanning data of the liquid article 20 with respect to the high energy X-rays and low energy X-rays; a controller 50 which controls each component of whole system so that they operate synchronously; and a computer data processor 60 for processing the data collected by the data collector and outputting inspection results.

[0091] As shown in FIG. 1, the radiation source 10 is placed at one side of the carrier mechanism 40 carrying the liquid article 20 to be inspected, while the detection and collection means 30 is placed at the other side of the carrier mechanism 40. The detection and collection means 30 comprises a detector and a data collector for acquiring the DR data and the multi-angle projection data of the liquid article 20. The data collector has a signal amplifying and shaping circuit, which operates under (current) integration mode or pulses (counting) mode. The detection and collection means 30 has its data output cable connected with the computer data processor 60 to store the collected data into the computer data processor 60 according to trigger instructions.

[0092] Besides, the inspection device also comprises a cylindrical article passage 20 made of metals and having openings at lower portions of side wall to allow the liquid article can be irradiated by radiations and shield some of the radiations that do not irradiate the liquid articles. The inspected liquid article 20 is placed in the article passage.

[0093] FIG. 2 shows a block diagram of the computer data processor 60 of FIG. 1. As shown in FIG. 2, the data collected by the data collector are stored in the memory 61 through an interface unit 68 and a bus 64. The configuration data and programs of the computer data processor are stored in the ROM (Read Only Memory) 62. The RAM (Random Access Memory) 63 is used for temporarily storing various data during the operating procedure of the processor 66. Besides, computer programs for data processing and a pre-compiled database are also stored in the memory 61, the database storing relevant information of various known liquid articles, such as names, kinds and physical attributes for comparing with the attributes such as density and atomic number of the inspected liquid article calculated by the processor 66. The internal bus 64 connects the memory 61, the ROM 62, the RAM 63, the input device 65, the processor 66, the display device 67 and the interface unit 68 together.

[0094] After the user inputs operation commands through the input device 65 such as keyboards and mouse, the instruction code of the computer programs will instruct the processor 66 to perform a predetermined data processing algorithm. After the processing results are obtained, they will be displayed on the display device 67 such as a LCD, or redirected in the form of a hard copy such as printing.

[0095] FIG. 3 shows a block diagram of the controller according to a first embodiment of the invention. As shown in FIG. 3, the controller 50 comprises a control unit 51 for controlling the radiation source 10, the carrier mechanism 40 and the detection and collection means 30 based on instructions from the computer 60; a trigger signal generation unit 52 for generating trigger commands for triggering the radiation source 10, the detection and collection means 30 and the carrier mechanism 40 to operate under the control of the control unit 51; a first driving motor 55 for driving the carrier mechanism 40 to ascend or descend according to the trigger command generated by the trigger signal generation unit 52 under the control of the control unit 51; a height information acquirement unit 53 for feeding the height information of the carrier mechanism back to the control unit 51 as the carrier mechanism movies; an angle information acquirement unit 54 for feeding the rotation angle of the carrier mechanism 40 back to the control unit 51 during the rotation process of the carrier mechanism 40.

[0096] According to the embodiment of the invention, the height information acquirement unit 53 and the angle information acquirement unit 54 both are photoelectrical coded discs, and thus they have the advantage of anti-interference.

[0097] As described above, in the database stored in the memory 61, identification information and physical attributes of various known liquid articles are stored in a tree structure. FIG. 4 shows a structure used for storing identification information and attribution information of various liquid articles in the database.

[0098] All the samples are divided into several subclasses, such as subclass 1 (alcohol), subclass 2 (cola), subclass 3 (milk), . . . , subclass n, and so on. Then, each subclass is subdivided into several subclasses. For example, subclass 1 (alcohol) is subdivided into subclass 1.1 (wine), subclass 1.2 (spirit), subclass 1.3 (beer), . . . , subclass 1.n and so on. Each subclass is further subdivided. For example, subclass 1.2 (spirit) is subdivided into subclass 1.2.1 (rum), subclass 1.2.2 (whisky), subclass 1.2.3 (vodka), . . . , subclass 1.2.n (Chinese spirit). A subclass is subdivided until the difference of the density and atomic number among respective samples in the subclass is smaller than a preset value, e.g. the system noise level. Such a subclass is a leaf node in the attribute database structure.

[0099] Furthermore, each leaf node is identified with all its father nodes' names, e.g. “40 percent rum of Havana, Cuba”. The identification corresponds to reference density and atomic number one by one. In the inspection process, the computer presents the tree structure to the user by levels, and the operator will input the identification information by levels. For example, the operator wish to acquire physical attribute of a bottle of 40 percent Havana Rum from Cuba, he/she can make selection by levels in the path of alcohol->spirit->rum->rum from Cuba->Havana Rum->40 percent.

[0100] The computer makes search as the human operator inputs identification level by level. When the user reaches the last identification, the corresponding reference density and atomic number are retrieved.

[0101] FIG. 5 is a diagram showing the relationship between a DR imaging and a CT imaging. According to the embodiment of the invention, a DR imaging is firstly performed on the liquid article to determine the liquid portion of the liquid article, and then a CT imaging is performed on the liquid portion only so as to improve the inspection speed.

[0102] FIG. 6 and FIG. 7 show examples of the result of a DR imaging, respectively. As shown in FIG. 6, after a DR imaging is performed on a liquid article, the liquid in the liquid article can be determined by analysis of pixels as described below. As shown in FIG. 6, the liquid article only contains a kind of liquid. However, as shown in FIG. 7, due to different attenuation coefficients of different kinds of liquid, when two or more kinds of liquid are contained in the liquid article and form several layers, the positions of the layer interfaces can be determined by analysis of pixels of the DR image obtained through a DR imaging. After that, CT imaging can be performed one by one on each layer of liquid.

[0103] FIG. 8 shows an outline flowchart of the liquid article inspection method according to the first embodiment of the invention. As shown in FIG. 8, liquid articles carried by a passenger should pass through security-inspection, for example, when he/she passes the customs. Firstly at step S110, the human operator places the liquid article to be inspected on the carrier mechanism 40, and acquires the identification information of the liquid article, e.g. 40 percent rum, from the customs declaration of the passenger or the remark of the liquid article.

[0104] Next, at step S111, the operator input the liquid identification into the database so that the system will acquire reference density and reference atomic number. Then, the operator presses a start bottom to start a DR scanning so as to generate a DR image, as shown in FIG. 6 and FIG. 7.

[0105] As described above, the purpose of DR scanning is to acquire a transmission image of the inspected liquid articles so that the operator can discern the internal structure of the inspected liquid articles to specify positions in the DR image where CT imaging should be performed. The system software can also use the DR image to automatically identify the positions of the liquid layers and guide the following CT imaging. The detailed process of the DR imaging will be described below.

[0106] What should be noted is that the DR scanning is optional. The CT scanning can be performed by directly specifying several positions without guidance of the DR scanning, to improve the inspection speed. For example, it is found that most liquid articles have at least 5 cm of liquid in height; so 5 cm from the bottom can be used as a pre-specified scan position. Furthermore, the operator can visually detect the size of the inspected article, and specifies a proper height experientially. For example, the scan height of canned Coca Cola can be set as 3 cm, while the scan height of a bottle of wine with a thick bottom can be set as 10 cm.

[0107] Having obtained the DR image, the CT scan positions can be determined either by automatic analysis of the DR image (step S113A), or by the operator using the input device 65 such as a mouse (step S113B), or by a mixture of both methods. In such a way, CT scanning is only performed at certain positions in the liquid articles, so that the inspection is speeded up without lowering the inspection quality.

[0108] Then CT scanning process is performed at step S114 at the determined positions in the liquid articles to obtain CT projection data and CT image are reconstructed based on the CT projection data. Each pixel of the CT image denotes the density, atomic number and other physical attributes of corresponding portion in the liquid articles.

[0109] Next, the computer analyzes the CT image by executing an analysis program, and obtains the measured density and atomic number at step S115. Then, at step S116, the measured density and atomic number are compared with reference density and atomic number retrieved from the database to determine whether they are consistent, such as whether the difference therebetween is smaller than a predetermined threshold. At step S117, the liquid article is indicated to be suspicious if the difference is larger than the predetermined threshold, and the operator is alarmed or the inspection result is printed.

[0110] Detailed operation of each step will be described with respect to FIGS. 9-14. FIG. 9 shows a flowchart of the process of a DR imaging, and FIG. 10 shows the arrangement of the DR image data collected by the detection and collection means 30 during the process of the DR imaging.

[0111] As shown in FIG. 9, during the DR imaging, at step S210, a command is sent to the controller 50 from the computer 60 to drive the carrier mechanism 40 to vertically move along the article passage 20. The controller 50 monitors the height of the carrier mechanism in real time through the height information acquirement unit 53 as the carrier mechanism vertically moves.

[0112] At step S211, the controller 50 sends a trigger signal to the detection and collection means 30 at intervals of a certain height (for example, 1 mm). The detection and collection means 30 receives the trigger signal, and then collects output signals from each detector to obtain high-energy detection data and low-energy detection data, and save them in its internal buffer.

[0113] At step S212, it is determined whether the carrier mechanism 40 reaches at a specified height or not, such as 500 mm. If not, then the flow proceeds to step S210.

[0114] If the carrier mechanism 40 reaches at the specified height, then the controller 50 will not send the trigger signal to the detection and collection means 30. The computer 60 reads collected high and low energy detection data from the detection and collection means 30, and arranges them in a matrix to form a DR image. Each pixel of the DR image records the residual intensity of the radiations after transmitting through the article, including low-energy radiation intensity and high-energy radiation intensity.

[0115] As described above, the CT scan positions are determined based on the DR image. By either automatic identification or manual specification, a row id in the DR image is obtained firstly, and then the id is converted to the height of the carrier mechanism by the computer, and the controller 50 is instructed to drive the carrier mechanism 40 to a specified position, and then a CT imaging is performed.

[0116] From the flowchart of the DR imaging, each row of the DR image corresponds to a height of the carrier mechanism 40. If it is assumed that the height of the carrier mechanism is 0 when the DR imaging starts, the carrier mechanism 40 descends during the imaging, and collection is trigged at intervals of h mm, then the m^th row in the DR image corresponds to a height of −m*h of the carrier mechanism.

[0117] FIG. 11 shows a flowchart of a process of determining a CT scan position by processing on the DR image.

[0118] In the DR image, the inspected liquid article is generally divided into a bottle bottom, a liquid portion, a bottle neck, a bottle cover and so on. The liquid portion can be extracted by an image analysis technique, and then the CT scan positions can be determined.

[0119] At step S310, a single-value DR image with low noise can be obtained by integrating and smoothing the high and low energy data of the DR image. For example, the specific method for integrating the high and low energy data can be selection of either the high or low energy data as the integration result, or weighted combination of the high and low energy data. The smoothing method can be a filtering process of the image with a Gauss filter.

[0120] At step S311, the liquid article (foreground) in the smoothed DR image is extracted, and the air (background) is removed. The specific method can set a threshold, and take pixels with values above the threshold as foreground pixels, and other pixels as background pixels. The principle of using a threshold to remove the background is that the liquid article blocks the radiations, and thus the corresponding pixels in the DR image have low values (the DR image records the residual intensity of the radiations).

[0121] At step S312, horizontal edge pixels in the smoothed DR image are extracted. The specific method is to calculate the difference between each pixel in the DR image and an adjacent pixel next in vertical direction, and take the pixel as the horizontal edge pixel if the difference is larger than a threshold.

[0122] At step S313, horizontal edge rows in the smoothed DR image are extracted. The horizontal edge rows are corresponding to the interface between the bottom and the liquid, the interface between the liquid and the air, and the interface between the cover and the air or interfaces in the container between different liquid layers. The method to get the horizontal edge rows is to calculate a ratio of the number of the horizontal edge pixels and the number of foreground pixels on each row, and take the row as the horizontal edge row if the ratio is larger than a threshold (for example, 50%).

[0123] At step S314, the DR image is divided vertically, and non-liquid regions are excluded. Horizontal edge rows in the DR image divide the DR image into a number of regions, including a bottle bottom, a liquid portion (may have several layers in different densities), a spacing portion inside the bottle (if any) and a bottle cover. The non-liquid regions can be excluded by establishing a choosing rule, which can be:

[0124] a) In the vertical direction, a region with the number of rows lower than a threshold is excluded. The region with a low number of rows is a region of little thickness, which might be the bottle bottom, the bottle cover or a spacing portion in the top of the bottle (for example, the air at the top of a can). The threshold can be set by investigating the bottle bottom, the bottle cover and the thickness of the air in the container of various containers of liquid packs.

[0125] b) In the horizontal direction, a region with average foreground pixel number of every row lower than a threshold is excluded. Such a region generally is the slender bottle neck. The threshold can be set by acquiring the width of the bottle necks of various containers of liquid packing.

[0126] At step S315, the CT scan position of the liquid region(s) is determined to locate respective layers of the liquid, excluding the non-liquid regions. The central rows in the height direction of these regions are taken as the CT scan positions.

[0127] Above described is the process for automatically determining the CT scan positions. In the case of manually specifying the scan positions, the operator directly specifies rows on the displayed DR image via the input device 65 as the CT scan positions.

[0128] FIG. 12 shows a process of a CT imaging, and FIG. 13 shows the arrangement of the CT projection data during the process of the CT imaging.

[0129] As shown in FIG. 12, after the CT positions have been determined, a CT imaging process is performed, i.e. a CT imaging is performed at the determined CT scan positions to generate a density-atomic number image of one slice of the inspected article, to measure the density and atomic number of the liquid. As described above, the CT imaging is only performed on the typical positions, and thus the passing time can be largely saved.

[0130] At step S410, the computer 60 sends a command to the controller 50 to drive the carrier mechanism 40 to rotate an angle, such as one degree. The controller 50 monitors the angle of the carrier mechanism in real time via the angle information acquirement unit 54 during the carrier mechanism rotates.

[0131] At step S411, the monitor 50 sends a trigger signal to the detection and collection means 30 after a rotation of one degree. The detection and collection means 30 receives the trigger signal, and collects output signals of respective detectors and saves them in its internal buffer.

[0132] Next, at step S412, it is determined whether cumulated rotation angle reaches a cycle or not. If not, then the flow proceeds to step S410 and continues the above process.

[0133] If the cumulated rotation angle reaches a specified angle (such as 360 degrees), then at step S413, the rotation ceases, and the controller 50 will not send a trigger signal to the detection and collection means 30 any more. The computer 60 reads out collected high and low energy detector signals from the detection and collection means 30 and arranges them in a data matrix to form CT projection data, as shown in FIG. 13. Each pixel of the CT projection data records the residual intensity of the radiations after transmitting through the article, including the low energy radiation intensity and the high energy radiation intensity.

[0134] At step S414, the computer 60 reconstructs a tomographic image of the density and atomic number, i.e. a CT image, from the high and low energy CT projection data by a dual-energy reconstruction algorithm. Each pixel of the CT image records the density and atomic number of the inspected article at the position corresponding to the pixel.

[0135] The process of reconstructing tomographic image from the high and low CR projection data will be described below.

[0136] Mathematical Principle of CT

[0137] A one-dimensional function p_θ(t) can be obtained by linearly integrating a two-dimensional distribution u(x,y) along a direction θ, which function is referred to as the projection of u(x,y) at an angle θ. If p_θ(t) of respective directions have been obtained, then the two-dimensional distribution u(x,y) can be accurately calculated through a Radon transform. The process of calculating a two-dimensional distribution from projection is referred to as reconstruction.

[0138] In real application, projection of attenuation coefficient of a slice of an article in respective directions can be measured by an X-ray machine and a detector rotating around the article a cycle. Then the two-dimensional distribution of the attenuation coefficient can be reconstructed from the CT principle.

[0139] Basis Material Decomposing Model

[0140] In the energy range of a mini X-ray security inspection system (<200 keV), the attenuation coefficient of radiation can be approximately expressed as following formula (1).

[0141] $\begin{array}{cc}\mu \left(E\right)=a_1{f}_p\left(E\right)+a_2{f}_{\mathrm{KN}}\left(E\right)& \left(1\right)\\ a_1=\frac{\rho Z}{M}{Z}^n& \left(2\right)\\ a_2=\frac{\rho Z}{M}& \left(3\right)\end{array}$

[0142] In formula (1), the linear attenuation coefficient, μ(E), as a function of X-ray energy E, is decomposed to f_p(E), which demotes the contribution from photoelectric effect, and f_KN(E), the Compton effect. Both f_p(E) and f_KN(E) have known formulas, which are omitted here. The decomposition coefficients a₁ and a₂ are related to the atomic number, mass number and density, with formulas being shown in (2) and (3), in which Z denotes the atomic number, M denotes the mass number, ρ denotes the density (g/cm³) and n is a constant.

[0143] According to formula (1), with a given X-ray energy distribution, the linear attenuation coefficient of each substance can be uniquely determined by only two coefficients a₁ and a₂. Therefore, if we select two basis materials, such as carbon and aluminum, then and all other materials can be expressed as the linear combination of the line attenuation coefficients of these basis materials, as shown in the following formula (4).
μ(E)=b₁μ₁(E)+b₂μ₂(E)  (4)

[0144] Formula (4) is just a linear transformation of formula (1), wherein μ₁(E) and μ₂(E) are the linear attenuation coefficients of the selected basis materials, and b₁ and b₂ are called basis material coefficients. Another interpretation of formula (4) is that the linear attenuation coefficient of any material can be regarded as a weighted sum of the linear attenuation coefficients of two basis materials.

[0145] Then we define a characteristic density, ρ*, as the product of the ratio of double atomic number and the mass number with the density, as in formula (5).

[0146] $\begin{array}{cc}{\rho }^{*}=\rho \frac{2Z}{M}& \left(5\right)\end{array}$

[0147] Assume that the atomic numbers and characteristic densities of the two basis materials are already known as (Z₁, ρ₁*) and (Z₂, ρ₂*) respectively, the atomic number and characteristic density of any material can be derived from the above formulas (1)˜(4) as follows.

[0148] $\begin{array}{cc}{\rho }^{*}=b_1{\rho }_1^{*}+b_2{\rho }_2^{*}& \left(6\right)\\ Z={\left(\frac{b_1{\rho }_1^{*}{Z}_1^n+b_2{\rho }_2^{*}{Z}_2^n}{b_1{\rho }_1^{*}+b_2{\rho }_2^{*}}\right)}^{1/n}& \left(7\right)\end{array}$

[0149] Basis Material Projection Model

[0150] The energy spectrum generated by an X-ray tube is typically a continuous spectrum. The energy response function of a detector to X-rays is not constant. Assume that the product of the energy spectrum N(E) and the energy response function P_d(E) is S(E), and S(E) is normalized as follows,

[0151] $\begin{array}{cc}{\int }_0^{E_m}S\left(E\right)dE=1& \left(8\right)\end{array}$
then the projection on a projection line can be expressed as following integration:

[0152] $\begin{array}{cc}p=-\ln \frac{I}{I_0}=-\ln {\int }_0^{E_m}S\left(E\right)\exp \left(-{\int }_l\mu \left(E,x,y\right)dl\right)dE& \left(9\right)\end{array}$
in which I₀ and I respectively denote the detector readings before and after the radiations are attenuated by the article, E_m denotes the maximum energy of the radiations, and l denotes the path of the radiations.

[0153] The above formula (9) shows the relationship between the measured projection p and the two-dimensional distribution μ(x, y). It is obvious that the formula (9) is not the linear integration of μ(x, y) because the X-ray energy is not a const, and thus does not meet the requirement of the mathematical principle of CT.

[0154] The conventional reconstruction algorithm neglects such uniformity, as a result the reconstructed image of μ(x, y) will have an artifact of a cup shape, which is called hardened artifact. If we calculate two sets of μ(x, y) by conventional reconstruction algorithm, and then derive information such as the atomic number and density, the result will also have artifacts.

[0155] The present invention solves this problem with basis material decomposition model. Substituting formula (4) in formula (9), we will get a basis material projection model,

[0156] $\begin{array}{cc}p=-\ln {\int }_0^{E_m}S\left(E\right)\exp \left(-{\int }_l^{}\left[{\mu }_1\left(E\right)b_1\left(x,y\right)+{\mu }_2\left(E\right)b_2\left(x,y\right)\right]dl\right)dE\mathrm{Let}& \left(10\right)\\ {\int }_l^{}{b}_1\left(x,y\right)dl=B_1& \left(11\right)\\ {\int }_l^{}{b}_2\left(x,y\right)dl=B_2& \left(12\right)\end{array}$
in which B₁ and B₂ are referred to as the projection of the basis material coefficients b₁(x, y) and b₂(x, y). Then the dual-energy projection data as follows can be obtained by collecting the projection data in dual-energy.

[0157] $\begin{array}{cc}{p}_1\left(B_1,B_2\right)=-\ln {\int }_0^{E_1}{S}_1\left(E\right)\exp \left[-B_1{\mu }_1\left(E\right)-B_2{\mu }_2\left(E\right)\right]dE& \left(13\right)\\ p_2\left(B_1,B_2\right)=-\ln {\int }_0^{E_2}{S}_2\left(E\right)\exp \left[-B_1{\mu }_1\left(E\right)-B_2{\mu }_2\left(E\right)\right]dE& \left(14\right)\end{array}$
in which E₁ denotes the maximum energy of the low-energy radiations, and E₂ denotes the maximum energy of the high-energy radiations. After (p₁, p₂) is measured, (B₁, B₂) can be solved based on the formulas (13) and (14), which will be described in the next section. And after (B₁, B₂) in all angles have been obtained, the distribution of the basis material coefficients b₁(x,y) and b₂(x,y) can be reconstructed according to the CT reconstruction theory. Then the atomic number and characteristic density distribution of the article and the linear attenuation coefficient of any energy can be calculated according to the basis material decomposition model.

[0158] Solution of Basis Material Coefficient Projection (B₁, B₂)

[0159] Both formulas (13) and (14) are logarithmic integral formulas, which can not be resolved analytically. The conventional non-linear iterative method needs a great deal of calculation, and can not easily obtain stable solutions.

[0160] It should be noted that the measured dual-energy projection can be expressed as follows after the radiations pass through the basis materials 1 and 2 with thickness of d₁ and d₂:

[0161] $\begin{array}{cc}{p}_1=-\ln {\int }_0^{E_1}{S}_1\left(E\right)\exp \left[-d_1{\mu }_1\left(E\right)-d_2{\mu }_2\left(E\right)\right]dE& \left(15\right)\\ p_2=-\ln {\int }_0^{E_2}{S}_2\left(E\right)\exp \left[-d_1{\mu }_1\left(E\right)-d_2{\mu }_2\left(E\right)\right]dE& \left(16\right)\end{array}$

[0162] Comparing formulas (13) and (14) with (15) and (16), it can be seen that the measured projection pair (p₁, p₂) is the same. That is, the projection data pair (B₁, B₂) of the basis materials is just the same as the thickness pair (d₁, d₂) of the basis materials. Therefore, the correspondence between the dual-energy projection data pair (p₁, p₂) and the basis material coefficient projection data pair (B₁, B₂) can be obtained by measuring the dual-energy projection of different thickness pair, and a look-up table can be formed. A pair (B₁, B₂) can be calculated from (p₁, p₂) according to the look-up table by linear interpolation, instead of a complex solution process.

[0163] FIG. 14 shows the process for measuring the attributes of the liquid.

[0164] As shown in FIG. 14, at step S510, the density image and the atomic number image are integrated and smoothed to form a single value CT image with low noise. The specific integration method can be selection of either the density image or the atomic number image as the integration result, or weighted combination of both images. The specific smoothing method can be filtering the image with a Gauss filter.

[0165] At step S511, the inspected articles (foreground, including the liquid and the pack thereof) in the smoothed CT image is extracted, and the air image (background) is removed. A specific method is to set a threshold, and take pixels with values above the threshold as foreground pixels, and other pixels as background pixels. The reason is that the density and atomic number of air are nearly zero, whereas those of the inspected liquid article are relatively larger.

[0166] At step S512, the liquid pixels in the foreground pixels are extracted. A specific method for extraction may include the following steps. Firstly, it establishes a binary image corresponding to the CT image, setting the value of the foreground pixels to one, the value of the background pixels to zero. Then the morphological erosion technique is applied to binary image to remove the packing, since the liquid is always inside the packing. The times of corrosions can be set in accordance with the thickness of the packing.

[0167] At step S513, the average density and average atomic number of all the liquid pixels in the CT image can be calculated as the output result of this measurement.

[0168] Furthermore, if the DR image analysis process finds that the liquid has multiple layers, the above steps are repeated with respect to each layer to determine if any layer is suspicious. The operator can be informed of the final inspection result.

[0169] Moreover, if the information of the liquid articles in the database is insufficient, ie., the system operator cannot find the identification information of a certain article, the database can be expanded. In other words, the database can not only be filled by the product manufacturer in the factory, but also be expanded by the operator on site. For example, when a new kind of drink appear in the market, the operator can generates reference density and reference atomic number from samples of the drink. FIG. 15 shows the process of expanding the database.

[0170] The basic procedure for adding an item to the database is to measure its reference density and atomic number, assign a unique identification to it, and then insert it in the identification tree with the reference density and atomic number.

[0171] As shown in FIG. 15, at step S610, the operator powers on the system, and logs in the database expansion interface. The system enters into the ready state after self check. Then the operator places liquid article to be inspected, which is desired to be added into the database, on the carrier mechanism 40. At step S611, the computer 60 sends a command to the controller 50 to trigger the radiation source 10 and the detection and collection means 30 to perform DR imaging. At step S612A, the position of the liquid is automatically determined as described above, or at step S612B, the position of the liquid can also be specified by the operator on the DR image.

[0172] Next, at step S615, the operator set identification for the samples of the liquid, such as Coca-Cola. At step S616, the identification is associated with reference density and atomic number and then stored in the database.

[0173] The above first embodiment relates to the case that a single liquid article is inspected at a time. A process of inspecting a plurality of liquid articles at a time will be described with respect to FIGS. 16-19. The second embodiment differs from the first embodiment in that the positions of imaging result displayed on the displayer should correspond to the positions of the articles on the carrier mechanism so that the human operator can ascertain which article is suspicious after the CT image has been obtained. FIG. 16A and FIG. 16B show diagrams of CT images reconstructed in the case that a plurality of liquid articles are to be inspected according to the second embodiment.

[0174] For example, if the operator observes the inspected articles on the carrier mechanism from the top, then the positions of respective articles on the CT images of all layers should correspond to the top view of the carrier mechanism.

[0175] FIG. 17A to FIG. 17K show the process how to rotate the CT reconstruction images and/or the carrier mechanism to be aligned with the position before CT scanning.

[0176] FIG. 17A shows a top view of the carrier mechanism 40 when the first CT imaging starts, in which the angle of the carrier mechanism is denoted by an arrow. FIG. 17B shows the first CT image, in which the rotation range of the carrier mechanism is identified by the dash line. Note that FIG. 17B is aligned to FIG. 17A by conventional CT reconstruction algorithm. FIG. 17C shows the first CT image displayed after the inspection, which is the same as FIG. 17B because no further rotation is needed.

[0177] FIG. 17D shows a top view of the carrier mechanism at the n^th CT imaging starts. From FIG. 17D it can be seen that the carrier mechanism rotates an angle with respect to that before the first CT imaging. FIG. 17E shows the CT image of FIG. 17D and FIG. 17F shows the n^th image displayed on the screen after inspection, which is aligned with the first CT image by rotation.

[0178] FIG. 17G shows a top view of the carrier mechanism at the last CT imaging starts. From FIG. 17G it can be seen that the carrier mechanism rotates an angle with respect to that at the first CT imaging starts. FIG. 17H shows the CT image of FIG. 17G, and FIG. 17I shows the last CT image displayed on the screen after inspection, which is aligned with the first layer of the CT image by rotation.

[0179] FIG. 17J shows a top view of the carrier mechanism after the last CT imaging ends, in which the carrier mechanism rotates an angle with respect to that before the first CT imaging starts. FIG. 17K shows a top view of the carrier mechanism after inspection, which returns to the position of FIG. 17A.

[0180] The basic alignment procedure is: after all the CT imaging, the angles of respective layers of the CT image and the carrier mechanism are adjusted. Firstly, respective layers of the CT images are rotated according to the angles of the carrier mechanism at respective CT imaging starts (the degrees can be obtained by the angle information acquirement unit 54) so that the positions of the same article in the CT images of respective layers are aligned, e.g. with the first layer of CT. Next, the angle of the carrier mechanism is adjusted so that the top view of the carrier mechanism corresponds to the CT image.

[0181] For example, assume that N times CT imaging are performed, and the angle of the carrier mechanism is α_n at the n^th CT imaging starts, and is β_n at the n^th CT imaging ends. The carrier mechanism rotates counterclockwise in the top view. In order for the position of the article in the n^th CT image to be consistent with that in the first one, the n^th CT image rotates by α_n-α₁ counterclockwise. And finally the carrier mechanism rotates by 360−(β_N-α₁) counterclockwise so as the top view of the carrier mechanism being consistent with the CT image.

[0182] FIG. 18 shows a flowchart for performing inspecting operation in the case of a plurality of articles are to be inspected. As shown in FIG. 18, at step S710, the human operator powers on the system and logs in the inspection interface. The system enters into the ready state after self check. Then the operator places the plurality of articles to be inspected, such as article A and article B, on the carrier mechanism 40 and presses an inspection button. Here, assume that the article A is placed at the up right corner of the carrier mechanism, and the article B is placed at the lower left corner of the carrier mechanism. Besides, the operator inputs respective identification information of article A and Article B.

[0183] At step S711, respective reference density and reference atomic number are retrieved from the database based on the identification information of article A and article B. Next, at step S712, the operator presses a start button to perform DR scanning as described above, to generate a DR image.

[0184] After the DR image has been obtained, at step S713A, the DR image is analyzed to automatically determine the scan position of the CT imaging, or at step S713B, the operator operates the input device 65 such as a mouse to specify positions where to perform a CT scanning. In such a way, CT scanning is only performed at typical positions in the liquid articles, whereby the inspection is speeded up without lowering the inspection quality.

[0185] Having determining the CT scan positions, a CT scanning process is performed at step S714, i.e. CT scanning is performed at the determined positions in the liquid articles to obtain CT projection data and a CT image is reconstructed based on the CT projection data. Each pixel of the CT image denotes the density and atomic number of corresponding portion in the liquid articles. In the case that the liquid is layered, the CT scanning is repeated for each layer.

[0186] After the last CT imaging ends, at step S715 and S716, the carrier mechanism and each layer of the CT image are adjusted as above so that the positions of respective articles in each layer of the CT image are consistent (e.g. aligned with those in the CT image of the first layer), and consistent with the real positions of the articles on the carrier mechanism (e.g. in the top view), thereby article B and article A can be distinguished from each other.

[0187] At step S717, image partition processing is preformed on the CT images of each layer (for example, by using a watershed algorithm) to obtain the liquid region of each inspected article. At step 718, the average density and average atomic number of respective pixels in each liquid region are calculated, and they are compared with reference density and atomic number at step S719. At step S720, it is determined whether the inspected article is suspicious.

[0188] At step S721, the results of respective layers are gathered and shown to the operator. One of the gathering methods is to conclude that the result is “secure” only if all the liquid regions in all the CT images are determined to be secure; otherwise the result is “suspicious”. Furthermore, the CT images of respective layers are colorized and displayed to the users. The suspicious articles are shown with a certain color (such as red), and secure liquid are shown with another color (such as green).

[0189] The case in which two liquid articles are inspected is described above. When more liquid articles are to be inspected, a plurality of regions, such as region A, region B, region C and region D shown in FIG. 19, are partitioned in the surface on which the carrier mechanism carries the liquid articles, as shown in FIG. 19, so that the operator can ascertain the positions of respective liquid articles. In this way, the operator can locate respective liquid articles in corresponding regions, and input respective liquid identification information for respective regions.

[0190] To improve inspection efficiency, and to help slim articles stand stably in the barrel, the third embodiment will employ a divided barrel.

[0191] The third embodiment differs from the second one in that a divided barrel is used in the process of inspecting a plurality of articles. The operation of the inspection system according to the third embodiment will be described with respect to FIGS. 20-25.

[0192] FIG. 20 shows a side view of a divided barrel according to an embodiment of the invention, FIG. 21 shows a top view of a divided barrel, and FIG. 22 shows a bottom view of a divided barrel.

[0193] As shown in FIG. 20, the divided barrel 70 comprises a bottom and a side wall coupled to the bottom. Protrudes of cone shape or other shape are provided on the lower surface of the bottom. The three cone-shaped protrudes can be inserted into the corresponding location holes on the carrier mechanism 40 so that the divided barrel 70 will rotate as the carrier mechanism 40 during the rotation of the carrier mechanism 40 to prevent relative motion from occurring therebetween.

[0194] Moreover, as shown in FIG. 20, a flange is provided along the top of the side wall to facilitate grasp and portage of the operator. The side wall has a sharp of a column or a cone, and is made of materials with elasticity, such as Polyethylene (PE) or aluminum.

[0195] FIG. 21 is the top view of three kinds of divided barrels. As shown, one or more dividing parts are provided in the space formed by the bottom and the side wall. The space is divided into plural subspaces to place corresponding liquid articles. In such a way, if a plurality of liquid articles is to be inspected at a time, the liquid articles are placed in the subspaces divided by the dividing parts. In this case, marks can be provided on the outer surface of the side wall to locate the articles in the barrel. For example, when one of four articles is determined to be suspicious, the one can be notified to the operator by the corresponding mark of the article on the side wall.

[0196] For example, as shown in left of FIG. 21, mark with round sections of preset sizes are provided at the end of the dividing parts, or one of the dividing parts being shorter than the others is used as the mark for locating liquid articles.

[0197] FIG. 22 is a bottom view of a divided barrel. Though three protrudes are provided uniformly at the lower surface of the bottom, the protrudes can also be distributed non-uniformly.

[0198] FIG. 23 shows a process how to automatically detect the divided barrel and the mark during an inspection operation. As described above, the divided barrel can be detected by a pattern matching method which is typical in the image processing because the divided barrel has a specific structure. Taking the first kind of barrel with a cross dividing part as an example, at step S810 a pattern image with a cross can be established at first, and the center of the pattern is placed on the center of the CT image to be identified to obtain a matching value.

[0199] At step S811, it is determined whether the matching value is larger than a predetermined threshold or not. If not, then the pattern image is rotated until maximum matching of the pattern image and the CT image is obtained. If the matching is larger than a predetermined threshold, then it is deemed that a divided barrel exists in the CT image; otherwise it is deemed that no divided barrel exists in the CT image.

[0200] In the case that a divided barrel exists in the CT image, then at step S812, the location mark can be detected according to the characteristics thereof. Again taking the first kind of barrel with a cross dividing part as an example, the location mark is at top of a dividing line which is longer than other three dividing lines. After it is determined that a divided barrel exists in the CT image through the pattern matching method, the cross line in the pattern image at maximum matching superposes a dividing line. The location mark can be detected by comparing the four dividing lines and taking the longest one.

[0201] In the case of a multi-layered CT imaging, firstly divided barrel detection and location mark detection are performed on each layer of the CT image. If no divided barrel is detected in respective layers, then it is deemed that no divided barrel is used by the operator. If a divided barrel is detected in at least one layer, then it is deemed that a divided barrel is used. If detected location mark positions of respective layers are different, then the one with strongest signal intensity can be taken as the final location mark. One method to describe the signal intensity of the location mark in a layer of the CT image is to subtract an average value from the maximum value of the four dividing lines in the layer. The larger the difference is, the stronger the signal intensity of the location mark is.

[0202] FIG. 24A to FIG. 24D shows a diagram of a process of rotating the barrel during the inspection operation. The divided barrel is adjusted to a specified position by adjusting the location mark of the barrel to a predetermined position. Taking the first kind of barrel with a cross dividing part as an example, a polar coordinate system is established with the center of the CT image as the origin, positions of each chamber are uniquely determined by the angle coordinates of the location mark. Assume that the angle of the polar coordinate corresponding to the finally determined location mark in the CT image is γ, and the preset adjustment target of the system is θ, then the adjustment can be done by rotating layers of the CT image and the carrier mechanism by θ-γ, as shown in FIG. 24 below.

[0203] As shown in FIGS. 24A and 24B, the divided barrel rotates a certain angle to reach a preset position. Similarly, each layer of the CT image rotates to be aligned with the rotated divided barrel.

[0204] FIG. 25 shows a flowchart of the inspection operation according to a third embodiment of the invention. Steps S910 to S921 are the same as steps S710 to S721 in the above second embodiment. Only steps S922 to S926 will be described.

[0205] At step S922, the CT image is analyzed. At step S923, it is determined whether there is a divided barrel. In the case of no divided barrel, then the inspection operation ends.

[0206] If a barrel is used, then at step S924, the location mark of the barrel is identified to determine the positions of each chamber. Next, at step S925, the carrier mechanism and each layer of the CT image are further adjusted so that each chamber of the barrel on the carrier mechanism and each chamber on each layer of the CT image reach predetermined positions.

[0207] At step S926, the system lists the results of respective liquid regions of each layer of the CT image in each chamber, and gathers them to display to the user. One gathering method according to the embodiment is to conclude that the result of a chamber is “secure” only if all the liquid regions of all the CT images in this chamber are determined to be secure; otherwise the result of the chamber is “suspicious”.

[0208] The above first to third embodiments describe the case where the inspection method according to the present inventions is used to determine whether the inspected liquid article is suspicious, but it can be directly determined whether drugs are concealed in the inspected liquid article.

[0209] The physical attributes (such as density and equivalent atomic number) of the liquid article will change after drugs are concealed therein. For example, the density of pure water is 1.00 g/cm³, and the atomic number thereof is 7.5.

[0210] After cocaine of 50 g is resolved in water of 1000 g, the density changes to 1.01 g/cm³, and the atomic number becomes 7.6.

[0211] The method for calculating the equivalent atomic number of a substance (including mixer) is as follows.

[0212] Assume that a substance includes N kinds of elements, and the atomic number of respective elements is Z_i, and the percent of the number of atoms of respective elements is α_i, wherein i=1,2, . . . , N and

[0213] $\sum _{i=1}^N{\alpha }_i=1,$
then the equivalent atomic number of this substance is:

[0214] $Z_{\mathrm{eff}}={\left[\sum _{i=1}^N{\alpha }_i{Z}_i^{4.5}/\sum _{i=1}^N{\alpha }_i{Z}_i\right]}^{\frac{1}{3.5}}$

[0215] Therefore, taking water (H₂O) as an example, the calculation of the equivalent atomic number is shown in Table 1 below.

[0216]

	TABLE 1
		kind of the atom
		H	O
	atomic number of	1	8
	respective atoms
	percent of number of atoms	66.7%	33.3%
	equivalent atomic number	7.51

[0217] After drugs are concealed therein, the change of the atomic number is shown in Table 2 below.

[0218]

	TABLE 2
	Without	With	With	With	With
	drugs	5% drugs	10% drugs	15% drugs	20% drugs
density	1.00	1.01	1.02	1.03	1.04
characteristic	1.11	1.12	1.13	1.14	1.15
density
atomic	7.51	7.62	7.73	7.84	7.95
number

[0219] Furthermore, FIG. 26A shows a change curve of the density after drugs are concealed in the liquid, FIG. 26B shows a change curve of the atomic number after drugs are concealed in the liquid, and FIG. 26C shows a change curve of the characteristic density after drugs are concealed in the liquid.

[0220] The inspection method according to the embodiment of the present invention will be described with respect to FIG. 27.

[0221] At step S1010, the operator places a liquid article to be inspected on the carrier mechanism 40, and acquires the identification information of the liquid article, e.g. a bottle of water, from the customs declaration of the passenger or the remark of the liquid article.

[0222] Next, at step S1011, the operator retrieves reference density and reference atomic number from the database based on the identification information, such as density of 1.00 (the characteristic density being 1.11) g/cm³, and reference atomic number of 7.51. After that, the operator presses a start button to start DR scanning, to generate a DR image.

[0223] Having obtained the DR image, at step S1013A, the DR image is analyzed to automatically determine CT scan positions, or at step S1013B, CT scan positions can be specified by the operator operating the input device 65 such as a mouse. In such a way, CT scanning is only performed at typical positions in the liquid articles, whereby the inspection is speeded up without lowering the inspection quality.

[0224] Having determining the CT scan positions, a CT scanning process is performed at step S1014, i.e. a CT scanning is performed at the determined positions in the liquid articles to obtain CT projection data and a CT image is reconstructed based on the CT projection data. Each pixel of the CT image denotes the density and atomic number of corresponding portion in the liquid articles.

[0225] Next, the computer analyzes the CT image by executing an analysis program, and obtains the measured density and atomic number at step S1015. For example, the density is 1.02 (characteristic density being 1.13) g/cm³, and the atomic number is 7.71. Then, at step S1016, difference of 0.02 g/cm³ between the measured density and reference density and difference of 0.20 between the measured atomic number and reference atomic number are acquired. If the drug concealment determination threshold is set as density difference of 0.01 g/cm³ and atomic number of 0.10, then the measured density and atomic number locate at the predetermined region. At step S1017, the liquid article is indicated to be suspicious if the difference is larger than the predetermined threshold, and the operator is alarmed or the inspection result is printed.

[0226] In order to determine reference density and atomic number, a sample of this kind of liquid is measured in advance and the result is stored in the database. In order to determine the desired threshold of the difference, it can be manually set to an appropriate value. If it is wished to detect minim drug, then the threshold can be set as a relative smaller value. The risk of doing so is misdetection which will occur to an inspected article without drugs due to system noise. On the contrary, if it is wished to reduce misdetection for articles without drugs, the threshold should be set as a larger value. The risk of doing so is non-detection which will occur to articles with minim drugs but the density difference being lower than the determination threshold.

[0227] As known to those skilled in the art, though the case where a single article is to be inspected is described above, the above fourth embodiment is also applicable in the case of multiple article inspection as the second embodiment and the third embodiment are.

[0228] Though the present invention is described with respect to the case of first DR imaging and then CT imaging, spiral CT imaging also can be adopted to inspect liquid articles according to the present invention.

[0229] A set of spiral CT images can be obtained by performing CT on liquid article. The position of the liquid in the liquid article can be determined by comparing and analyzing pixels in the set of CT images, and whether the liquid is layered can also be determined. The physical attributes, such as density and atomic number, of the liquid of respective positions can be obtained in a similar manner as described above. For example, when spiral CT imaging is performed on the liquid article shown in FIG. 28, and the spiral pitch is 2 cm, then a set of CT images as shown in FIGS. 29A-29M can be obtained. In this way, the position of the liquid in the liquid article can be obtained by analyzing pixels in the spiral CT images. Here, the spiral CT imaging can be high pitch CT imaging or normal pitch CT imaging.

[0230] Moreover, though the above description takes the density and atomic number as examples, the present invention is also effective when only one attribute, either the density or atomic number, is used, or even more physical attributes are used to identify suspicious articles.

[0231] Though in the above description first the DR imaging and then dual-energy CT imaging are performed to acquire the density and atomic number of the liquid, the DR imaging is optional. Predetermined positions where to perform dual-energy CT imaging can be specified in advance for various liquid articles to acquire the density and atomic number of the liquid.

[0232] While exemplary embodiments of the present invention have been described hereinabove, it should be clear to those skilled in the art that any variations and/or modifications of the basic inventive concepts will still fall within the scope of the present invention, as defined in the appended claims.