US 4,498,104 AGrant
Method and Apparatus for Concealing Errors in a Digital Television Signal
Issue Date:1985-02-05
•6 Claims
•3 Drawing Sheets
Abstract
The digital signal values of adjacent picture points of pairs of points straddling the picture point currently being processed in every direction are compared in terms of only their more significant bits to determine which, if any, of the pairs involve a difference of value greater than a predetermined threshold. The determination that such a value difference is present is treated as the recognition of a picture contrast feature and causes, through operation of a read-only memory, the exclusion of such a picture point pair from the averaging process for producing a substitute value for the picture point currently under processing when the latter is found unusable by means of an error-detector. The resulting system for concealing errors detected in television picture avoids disturbing the outlines of picture features as other systems do which substitute averages of neighboring picture points for picture points found to be erroneous.
Metadata
Assignee
- Robert Bosch GmbH
Inventor
- Axel Schulz
Application Information
Application Number:US 3823857
Filing Date:1982-05-26
Priority Date:1981-05-30
Art Unit:262
Classifications
IPC:
H04N 514
Field of Search:
358382360160;163;166;21 R;37;314;336;16752;5438.1
Patent Drawings (3 sheets)
Description
This invention concerns a method and apparatus for substituting another digital signal for a digital television signal that has been found to be erroneous, the substituted signal being derived from digital signals relating to neighboring picture points.
Methods and systems for making it instantly possible, when an erroneous picture point signal is detected, to substitute the average value of two neighboring picture points on a line passing through the picture point represented by the erroneous signal are disclosed in two copending patent applications owned by the assignee of this application, namely Ser. No. 357,619, filed Mar. 12, 1982 and Ser. No. 366,936, filed Apr. 9, 1982, now U.S. Pat. No. 4,464,674, regarding each of which I am a joint applicant. The methods and systems there described proceed on the basis of the recognition that television pictures, so far as the brightness information is concerned, have a considerably redundant structure, and the value of neighboring picture points correspond more or less to the value of an unspoiled picture point. At edge transitions, however, for example where a jump from black to white, or vice versa, takes place in a television picture, the methods described in these prior applications are open to serious error, since neither a value representing white, nor a value representing black, will in general be substituted for the defective picture point signal, but rather a gray value.
The Invention
It is an object of the present invention to make it possible quickly and effectively to determine whether a detected erroneous picture point signal lies at an edge between high and low brightness value and, if it does, to substitute a value derived from neighboring picture points which will not disturb the representation of that edge.
Briefly, the various pairs of picture points available for providing a derived value are examined for the presence of a steep change in brightness value and when a steep change is detected, picture points recognizing the picture structure thus revealed are used to provide a substituted value for the erroneous value. The same system of delay circuits delaying signals by one line, and in each case also making available several signals for each line progressively delayed by one picture point, that is described in the aforesaid copending applications, is utilized together with additional circuitry necessary for operation in accordance with the present invention.
The present invention has the advantage that the analysis of the picture content in the vicinity of a disturbed picture point signal excludes neighboring picture point signals showing large value differences from the process of deriving a new picture point value for substitution for the erroneous signal. This excludes a source of error to which the earlier systems were vulnerable.
There is also the advantage that existing circuit units can be modified in a simple way without considerable additional expense for the practice of the invention. Furthermore, by the provision of hysteresis in the comparison of neighboring picture point values straddling a disturbed picture point, certain deviations of the picture point values can be allowed.
Brief Description of the Drawings
The invention is further described by way of illustrative example with reference to the annexed drawings, in which:
FIG. 1 is a diagram of picture points of a portion of a television picture where a horizontal black-white transition takes place;
FIG. 2 is a block diagram of a circuit according to the invention, and
FIG. 3 is a diagrammatic representation of a comparator system having hysteresis utilized in the circuit of FIG. 2.
Description of the Illustrated Embodiments
Corresponding portions of five successive television lines designated n-2 to n+2 are shown in FIG. 1, each line consisting of a number of picture points forming part of the point raster of the television picture. Of the five lines the middle line n is intended to represent a line currently being reproduced, while the line n-1 is the line preceding it, the line n-2 the next but one preceding it, the line n+1 the line following the line n and the line n+2 the line following n+1. Seven picture points of each line are shown and the point symbols indicate that the line n-2, n-1 and n contain white picture points and the line n+1 and n+2 black picture points. A picture point of line n designated by a cross represents a picture point disturbed by drop-out during the reproduction from a magnetic tape in a digital recording and reproduction apparatus.
In accordance with the procedures described in the aforesaid prior applications, a new value is derived from the values of neighboring picture points around the disturbed picture point which is then substituted in place of the disturbed picture point. From the brightness jump or step running horizontally in FIG. 1 and providing a black-to-white transmission between lines n and n+1, carrying over the neighboring point values to provide a substitute value for the disturbed point would derive a medium gray value from the vertically neighboring picture points a and b, which would disturb the luminance structure of the black-to-white transition when substituted in place of the disturbed picture point.
According to the present invention an analysis is made of the picture content in the area surrounding the disturbed picture point x and on the basis of this analysis, as the result of the recognition of a larger luminance step or jump, an error processor is interposed for calculating the new picture point value. The error processor is so controlled by a control circuit that, for example, the picture points a and b are excluded from the average value calculation in the case of the situation illustrated in FIG. 1 and, for example, the picture points c and d are utilized for providing an average value.
As shown in FIG. 2, a series-to-parallel converter 21 is connected at the output of a magnetic tape apparatus 20 used for storing digitally-coded color television signals on a tape. The digital video signal at the output of the series-to-parallel converter 21 which appears in 8-bit parallel serial representation of the successive picture points is then supplied successively to two delay units 22 and 23 each of which delays the signal by the time period of one television line. The signal at the input of the delay circuit 22 represents the signal of the line n+1 in terms of the diagram of FIG. 1, the signal at the output of the delay unit 22 represents the currently-reproduced line n and the signal at the output of the second delay unit 23 represents the signal of the line n-1 that follows the currently reproduced line.
The signal of the line n+1 is supplied to a first shift register device 24, the three outputs of which respectively provide successive picture points simultaneously by virtue of a delay of one picture point interval in each cell of the shift register. Thus, at the first output of the shift register 24, the signal available represents the picture point a+, the signal at the second output represents the picture point a and the one at the third output of this shift register 24 represents the picture point a-. In a similar manner the digital video signals of the currently processed line n are supplied to a second shift register device 25 from the output of which the picture points c, x and d are made available. Finally, the digital video signals of the line n-1 are supplied to a third shift register device 26, at the output of which three successive picture points of the line n-1 are provided, namely b+, b and b-. The signals provided simultaneously at the outputs of the shift register devices 24, 25 and 26 thus provide, in addition to the picture point x, which is assumed to be erroneous, the immediately adjacent picture point surrounding the picture point x, as well in the horizontal and vertical direction as in the directions of the two diagonals.
The signals at the outputs of the shift registers 24, 25 and 26, which appear in serial 8-bit-parallel form are all, with the exception of the signal of the currently-processed picture point x, supplied to the 8 multibit inputs of an error processor 30 mentioned in the copending patent applications. At the same time these signals are supplied in pairs to four circuits 31, 32, 33 and 34 which are 4-bit magnitude comparators, as shown in FIG. 3. In this manner the signals corresponding to a pair of picture points diametrically straddling the currently-processed picture point x are in every case provided to the inputs of a comparator. Thus, the signals from the middle outputs of the respective shift register units 24 and 26 provide the signal pair a/b, the signals from the first output of the shift register 24 and from the third output of the shift register 26 provide the signal pair a+/b-, the signals from the third output of the shift register 24 and the first output of the shift register 26 form the signal pair a-/b+ and, finally, the two signals from the first and third outputs of the shift register 25 form the signal pair c/d. All the signal pairs are subjected to a comparison treatment in one of the comparators 31 to 34 explained below in connection with FIG. 3. The signals at the outputs of the comparators 31, 32, 33 and 34 serve as addresses for a read-only memory 35 to extract control signals therefrom for control of the error processor 30.
For the recognition of disturbed picture points an error recognition circuit can be built into the serial-to-parallel converter 21 which in case of detection of a disturbance provides an output signal consisting of 8 bits having a value which is intended as data information relating to the gray scale in the video signal. For example, a disturbed picture point can be designated by the value 0000 0000. The gates 36, 37 and 38 at the inputs and outputs of the delay units 22 and 23 then respond to the recognition of the occurence of a disturbed picture point provided at the output of the series-to-parallel converter 21, respectively at once, after a delay of one line and after a delay of two lines. After series-to-parallel conversion in series-to-parallel converters 39, 40 and 41 these signals are supplied to the read-only memory 35. The circuit can also be constituted in such a manner that an error line runs from the error recognition circuit incorporated in the series-to-parallel converter 21 and provides a signal in response to the occurence of a disturbed picture point, at once to the gate 36 and after corresponding delays to the gates 37 and 38. Each s/p converter supplies, in parallel, one bit for each of three picture points, each bit signifying whether or not the picture point is disturbed. These gates and s/p converters provide to the ROM 35 a basis for causing the exclusion from further processing of a pair of picture points (bridging the point x) containing at least one disturbed picture point.
The comparators 31, 32, 33 and 34 are provided for selection of the undisturbed picture points located in the neighborhood of a disturbed picture point for forming an average value to be substituted for the disturbed picture point value, as already above explained. Such a comparator is shown with its input and output connections in more detail in FIG. 3. At the respective inputs A and B it can be seen that 8-bit parallel data signals are serially supplied there to the comparator 31. In order that in comparison of the values of the two signals used for replacing the disturbed picture point minor differences between the signals should not interfere with the provision of an average value, the comparator inputs are so constituted that only the bits of the 8-bit parallel signal which correspond to the higher value steps are supplied to the inputs A and B, while the bits of lower value are disregarded or suppressed. In the illustrated example the four more significant bits are used. This provides a desired hysteresis in the comparison of the two signals compared, in the sense that only the four most significant bits of the proposed substitution signals are compared with each other. In other words, the difference must exceed a predetermined value in order for the comparison result to be considered as a basis for preferring one pair of points over another pair. The comparator itself is constituted as an integrated circuit, for example, the unit available from Texas Instruments with the designation 74 LS 85. The three outputs of the comparator 31 and, according to the result of the comparison, provide a signal indicating either that the value of the signal at the input A is smaller than the one at the input B, or that the signal at the input A is greater than the signal at the input B, or else, that within the limits of the comparison conditions the signal A is actually or practically equal to the signal at the input B.
In accordance with the result of the comparison the read-only memory 35 (FIG. 2) is caused to respond and to select the picture points which have the greatest degree of agreement in value within the pair, or also of the pairs compared with each other, for the formation of an average value. In this manner it is assured that a picture portion having a relatively large black-to-white step in which a disturbed picture point appears will not result in the disturbed picture point being undesirably replaced by a gray value. The content of the read-only-memory 35 is similar to the ROM in the copending patent applications.
It is evident that two pairs of picture points may have the same intra-pair difference, as measured by the comparators 31-34, and in the case of still many point pairs the differences will be below the threshold of the four most significant bits. The order of selection of equally suitable pairs of picture points, so far as the comparators 31-34 may have failed to distinguish any basis of preference, can conveniently be determined in any desired way in the processor 30. For example, if averaging in a horizontal direction is preferred, the horizontal pair can be given first priority in the order of selection followed by one diagonal, the other diagonal and, last, the vertical pair. The ROM 35, not only serves to exclude from selection picture point pairs containing a disturbed picture point, but also, in cooperation with the processor 30, operates as control means for selecting the least differing picture point pair, and even if two or more pairs have the same difference or have differences below the threshold of the comparators 31-34, will determine the pair of which the average value will be used to substitute for the value of a disturbed picture point.
It should be further mentioned that whenever there is no disturbance present in the video signal, the picture point signals simply run through the circuit of FIG. 2 without any kind of manipulation like in the copending applications. Only when and if it is recognized that as the result of testing the very picture point signal that is currently being processed, an error or disturbance is present, is a particular address pattern applied by the gates 36, 37 and 38 to the ROM 35, which then responds, taking account of the values of the neighboring nonfaulty picture point pairs as analyzed by the comparators 31-34 exercises an appropriate control on the error processor 30 and causes a particular value to be computed from the neighboring picture points, drawing into account more or less neighboring picture points for this average value formation.
It should also be stressed that in the explanation of the invention in particular regard to the block circuit diagram of FIG. 2, no account has been taken with regard to the timing requirements of the circuit technology processing the signal, i.e. nothing has been said regarding a certain time sequence of the individual signals and their replacement by other signals. Propagation time differences as the result of different circuit treatment of disturbed and undisturbed signals have not been taken into account, but it should be understood that the conventional methods of timing circuits for processing digital signals and providing logic operations are usable in the circuits here referred to in essentially the same way as in other digital signal and logic circuits. It is well-known how to provide, either by the insertion of very small delays or by the use of clocking circuits, for avoiding malfunction of circuits because of timing problems.
Although the invention has been described with reference to a particular illustrative example it will be understood that modifications and variations are possible within the inventive concept.
A further advantage is performed of assigning different priorities to different picture point pairs with reference to their positions with respect to other undisturbed picture points.
Claims
I claim:
1. Method of concealing errors in a digital video signal comprising the steps of: making available simultaneously by a signal delay network the value of a digital signal representing a first picture point currently being reproduced and signals representing all the picture points immediately adjacent to said first picture point; subjecting said signal representing said first picture point to an error detecting test; at least in the event that said error detecting test detects the presence of error in said signal representing said first picture point, comparing the signal values representing the members of each and every pair adjacently straddling said first picture point, determining in each case whether the members of a pair of said signal values differ by more than a predetermined amount, and providing comparison result signals representative of said determinations; in the event that said error detecting test detects the presence of an error in said signal representing said first picture point, causing the exclusion from further processing regarding said first picture point of all signals representing picture point pairs found by said comparison and determination step to differ from each other by more than said predetermined amount and then causing the values of signals representative of at least one other pair of said picture points to be averaged together to produce an average value signal, and inserting said average value signal in an output digital video signal stream in place of said digital signal representing said first picture point.
2. Method as defined in claim 1 in which the step is also performed of assigning relative priorities to different picture point pairs with reference to their position with respect to said first picture point.
3. Method as defined in claim 1 in which said picture point pairs, the signals representing which are not excluded from processing regarding said first picture point, and for which pairs said comparing step had provided identical comparison results, are respectively selectable for forming an average value according to a predetermined priority scheme based on their position with respect to said first picture point.
4. Apparatus for concealing errors in a digital video signal comprising the steps of: means for detecting the presence of an error in any combination of digital signals representing a first picture point among said video signals; a network of delay circuits for simultaneously making available at respective outputs of said network, as each picture point signal successively becomes said first picture point and is, in its turn, being processed for error concealment, said first picture point signal being currently processed and also a multiplicity of other picture point signals representing all pairs of picture points bridgingly adjacent to said first picture point for contributing to the processing of said first picture point; means for comparing with each other the members of each said pair of said picture point signals simultaneously made available with said picture point signal of said first picture point and thereby determining in each comparison whether the members of the pair of picture point signals differ from each other by more than a predetermined amount of picture point brightness value; means for forming an average value in response to the values of the picture point signals constituting at least one selected pair of picture point signals; means for excluding picture point signals of picture point pairs found by said detecting means to contain at least one disturbed picture point from selection for averaging in said averaging means, and control means responsive to the detection of an error by said error-detecting means, in said video signal representing said first picture point, for selecting paired picture point signals, at least partly on the basis of minimizing intra-pair differences as detected by said comparing means, from said picture point signals not excluded by said excluding means, for averaging in said averaging means and for substituting the result of said averaging means for the video signal originally representing said first picture point.
5. Apparatus as defined in claim 4 in which said exclusion means is constituted by a read-only memory (35) having its address input connected with the outputs of said comparison and determination means and having its outputs connected for control of said averaging means.
6. Apparatus as defined in claim 4 in which said comparison and determining means has the said predetermined value fixed by the number of the more significant bits of the picture point video signals that are admitted to the input of said comparison means.
Patent Citations (5)
Non-Patent Literature (2)
- Journal of the SMPTE; "Digital Television Image Enhancement", by John Rossi, vol. 84, pp. 545-551, (1975).
- "Digital Techniques for Reducing Television Noise" by John P. Rossi, SMPTE Journal, Mar. 1978, vol. 87, pp. 134-140.