Image Reading Apparatus for Correcting Density Levels

1. Field of the Invention

The present invention relates to an image reading apparatus for scanning and exposing an original image, and photoelectrically converting the image using an image sensing element and, more particularly, to an apparatus for reading a tone image such as a photograph.

2. Related Background Art

Conventional image reading apparatuses using image sensing elements such as CCDs have been available.

FIG. 1 is a block diagram of an image reading apparatus of this type. In FIG. 1, light reflected by, e.g., an original placed on an original table, or projection light through a microfilm is scanned and photoelectrically converted by an image sensing unit 61 using an image sensing element such as a CCD, and an electrical signal is quantized into multi-value data by an A/D converter unit 62.

The multi-value quantized data is supplied to the next tone processing unit 63, and is converted by, e.g., a dither method, into binary data (e.g., 1/0) which allows reproduction of a tone image. The binary data is output from the reading apparatus.

The arrangement shown in FIG. 1 faithfully reads an original image having a low contrast. Therefore, output data is reproduced to have a low contrast as in the original image, thus posing a problem.

As shown in FIG. 2, another arrangement has been proposed wherein a .gamma. correction unit 64 is arranged between the A/D converter unit 62 and the tone processing unit 63.

In FIG. 2, multi-value quantized data output from the A/D converter unit 62 is input to the next .gamma. correction unit 9 to correct a difference between photoelectric conversion characteristics of the image sensing element such as a CCD and human tone visual recognition characteristics. FIGS. 3, 4, and 5 show this state.

FIG. 3 represents photoelectric conversion characteristics of the image sensing element. In FIG. 3, the abscissa corresponds to a scale of density which linearly changes with respect to human tone visual recognition characteristics. In this manner, since the photoelectric conversion characteristics of the image sensing element are normally different from the human tone visual recognition characteristics, a .gamma. correction table 65 shown in FIG. 4 is used to perform correction (.gamma. correction), so that an output D' linearly changes with respect to the human tone visual recognition characteristics, as shown in FIG. 5.

The .gamma.-corrected signal is supplied to the next tone processing unit 63, and is converted by, e.g., a dither method, into binary data (e.g., 1/0) which allows reproduction of a tone image, as shown in FIG. 2. The binary data is output from the reading apparatus.

The arrangement shown in FIG. 2 faithfully outputs an original image having a low contrast. Therefore, output data is reproduced to have a low contrast as in the original image, thus posing a problem.

As shown in FIG. 6, still another arrangement which has a tone processing unit 67 using a dither method to reproduce a pseudo halftone image on the basis of multi-value quantized data from the A/D converter unit 62 has been proposed.

Multi-value quantized data obtained by multi-value quantization by the A/D converter unit 62 is supplied to the tone processing unit 67, and is converted by, e.g., a dither method, into binary data (e.g., 1/0) which allows reproduction of a tone image. More specifically, dot patterns having dot densities proportional to density data of an image are sequentially read out from a dither pattern memory 66 storing dot patterns shown in FIG. 7, and the readout patterns are output.

However, a tone of an original whose image entirely has a low density and a low contrast, as shown in FIG. 39, or of an original whose image entirely has a high density and a low contrast, as shown in FIG. 40 is faithfully reproduced. Therefore, a reproduced image has a low contrast as in the original image.

The above-mentioned problems also occur when an image of a normal document, book, or the like is read, and are conspicuous when an image recorded on a film is to be read.

A density and a contrast of an image recorded on a film largely depend on various conditions such as photographing conditions, development conditions, and the like as well as a type of film. Therefore, this implies that images have quite different densities and contrasts in units of films to be read.

Therefore, when an image recorded on a film is to be read, reading cannot be satisfactorily attained unless the above-mentioned problems are solved.

The present invention has been made in consideration of the above situation, and has as its object to satisfactorily read an image having an unbalanced density distribution, e.g., an image having an extremely higher or lower density level as a whole and an image having a low contrast.

It is another object of the present invention to provide an image reading apparatus which can obtain a high-contrast image from a low-contrast original image.

It is still another object of the present invention to provide an image reading apparatus which can obtain a high-contrast tone image from a low-contrast original image without impairing an original tone.

It is still another object of the present invention to provide an image reading apparatus suitable for reading an image recorded on a film.

The above and other objects and features of the present invention will be apparent from the following description.

FIGS. 1, 2, and 6 are block diagrams showing arrangements of conventional image reading apparatuses;

FIGS. 3, 4, and 5 are graphs showing conversion operations by the arrangement shown in FIG. 2;

FIG. 7 shows dither patterns;

FIG. 8 is a graph showing a conversion operation by the arrangement shown in FIG. 6;

FIG. 9 is a block diagram showing the first embodiment of the present invention;

FIGS. 10 to 15 are graphs showing operations according to the first embodiment of the present invention;

FIGS. 16 and 17 are graphs showing operations according to the second embodiment of the present invention;

FIG. 18 is a view showing the third embodiment of the present invention;

FIG. 19 is a block diagram of the fourth embodiment of the present invention;

FIGS. 20A to 20C and 22A to 22C are graphs for explaining tone correction according to the fourth embodiment;

FIGS. 21 and 23 show density histograms representing the effect of the fourth embodiment;

FIGS. 24A to 24C are graphs for explaining tone correction according to the fifth embodiment of the present invention;

FIG. 25 is a view for explaining an original reading range;

FIG. 26 is a block diagram showing the seventh embodiment of the present invention;

FIGS. 27, 29, 30, and 32 show density histograms showing frequencies of appearance of density levels of images;

FIGS. 28 and 31 show dither patterns corresponding to the density histograms;

FIG. 33 shows an image obtained by converting an image with dither patterns;

FIG. 34 is a block diagram showing the eighth embodiment of the present invention;

FIG. 35 is a graph showing an I/O relationship of a .gamma. conversion unit;

FIG. 36 is a graph for explaining a method of determining maximum and minimum densities D.sub.max and D.sub.min according to the ninth embodiment of the present invention;

FIG. 37 shows a dither pattern according to the ninth embodiment of the present invention;

FIG. 38 shows a region on an original from which frequencies of appearance of density levels are detected according to the tenth embodiment of the present invention; and

FIGS. 39 and 40 show images.

The preferred embodiments of the present invention will be described in detail hereinafter.

First Embodiment

FIG. 9 is a block diagram of the first embodiment of the present invention. In the embodiment shown in FIG. 9, a scanning operation for detecting frequencies of appearance of image densities (to be referred to as prescanning operation hereinafter) is performed prior to a scanning operation for outputting image data.

Control for scanning an original is performed by a CPU 11 by driving a scanning system and an exposure system (neither are shown) for guiding image light from an original to an image sensing unit 1. Scanning and exposure arrangements and control are known to those who are skilled in the art, and a detailed description thereof will be omitted.

An analog image signal obtained by photoelectrically converting image light from an original by the image sensing unit 1 including an image sensing element such as a CCD is quantized into multi-value data by an A/D converter unit 2. Note that the A/D converter unit 2 outputs white level as a value "0", and outputs black level as a value "255". In the prescanning mode, a switch 6 is set at an adder 7 side under the control of the CPU 11.

The multi-value quantized data is input to the adder 7 to count frequencies of appearance in units of densities (to be referred to as a density histogram hereinafter). The count value is stored in a histogram memory 8. FIGS. 10 and 13 show contents of the histogram memory 8. FIG. 10 shows a density histogram of an image which entirely has a low density and a low contrast. FIG. 13 shows a density histogram of an image which entirely has a high density and a low contrast.

Upon completion of the prescanning operation, the CPU 11 looks up the content of the histogram memory 8, and outputs a conversion table to a conversion table memory 10. FIGS. 11 and 14 show this state.

More specifically, as shown in FIGS. 10 and 13, if minimum and maximum densities appearing in an original are respectively represented by D.sub.min and D.sub.max, the CPU 11 looks up the histogram memory 8 to detect these D.sub.min and D.sub.max, and calculates a conversion table for performing the following three conversions according to the relationship among a value of output data D from the A/D converter unit 2, D.sub.min, and D.sub.max.

The CPU 11 calculates a D-D' conversion table based on equations [1] (see FIG. 11 or 14), and writes the calculation results in the conversion table memory 10.

The CPU 11 then switches the switch 6 to a converter 9 side to perform a scanning operation for outputting an image signal. In this case, the multi-value quantized data D from the A/D converter unit 2 is input to a converter 9, and is level-converted to data D' for changing a level between 0 to 255 according to the conversion table stored in the conversion table memory 10. Thereafter, the data D' is subjected to tone processing such as dither processing by a tone processing unit 5, and the processed data is then output.

As a result, an image having the density histogram shown in FIG. 10 is level-converted according to the conversion table shown in FIG. 11, or an image having the density histogram shown in FIG. 13 is level-converted according to the conversion table shown in FIG. 14. Therefore, each of these image data is output as data D' as if it originally had a high-contrast density histogram having density levels falling within a range of 0 to 255, as shown in FIG. 12 or 15. That is, a high-contrast image can always be obtained from a low-contrast image.

Second Embodiment

In the second embodiment, the CPU 11 detects D.sub.max and D.sub.min of a density whose frequency of appearance exceeds a predetermined value (N) from a density histogram generated by the prescanning operation like in the first embodiment, calculates a conversion table according to equations [1] using the detected densities, and outputs the calculated conversion table to the conversion table memory 10. FIGS. 16 and 17 show this state.

Thereafter, like in the first embodiment, level conversion is performed according to the conversion table stored in the conversion table memory 10, and tone processing such as dither processing is performed by the tone processing unit 5. In this embodiment, densities whose frequencies of appearance are below the predetermined value (N) are ignored in order to avoid level conversion errors caused by dust or scratches on an original. As the predetermined value (N), an appropriate amount is used according to a reading resolution of the image sensing unit 1 (e.g., a data volume per mm.sup.2), and a minimum size to be read as a highest (or lowest) density region. This embodiment is effective to prevent level conversion from being influenced by dust or scratches when an enlarged projection image of, e.g., a microfilm is to be read.

Third Embodiment

In the third embodiment, when a density histogram is generated by the prescanning operation like in the first embodiment, a photometric region 12 is set at the central portion, and frequencies of appearance are counted in this region instead of counting frequencies of appearance of an entire reading region 13, as shown in FIG. 18.

A means for attaining this operation is realized by regulating a driving portion of a sensor such as a CCD included in the image sensing unit during the prescanning operation or by limiting a data input period to the adder 7 under the control of the CPU 11.

The CPU 11 looks up a density histogram generated for the central photometric region 12, and detects maximum and minimum densities D.sub.max and D.sub.min (or D.sub.max and D.sub.min for a density whose frequency of appearance exceeds a predetermined value as in the second embodiment). The CPU 11 then outputs a conversion table based on equations [1] using the detected D.sub.max and D.sub.min to the conversion table memory 10 like in the first embodiment. Thereafter, like in the first embodiment, level conversion is performed according to the conversion table stored in the conversion table memory 10, and tone processing such as dither processing is performed by the tone processing unit 5.

In this embodiment, when an original has a small size and is present not in the entire image to be read by the image sensing unit 1, as shown in FIG. 18 (in FIG. 18, 14 denotes an original), a level conversion error caused by a portion 15 having no original in the image to be read can be avoided.

As described above, according to the first to third embodiments, since a correspondence in conversion of a level conversion means is changed according to frequencies of appearance in units of density levels of an original, a high-contrast tone image can be obtained from a low-contrast original without impairing an original tone.

Fourth Embodiment

FIG. 19 is a block diagram of the fourth embodiment. Note that the same reference numerals in FIG. 19 denote the same parts as in FIG. 9.

In this embodiment, a scanning operation for detecting frequencies of appearance of image densities (to be referred to as prescanning operation hereinafter) is performed prior to a scanning operation for outputting an image like in the first embodiment.

Control for scanning an original is performed by a CPU 11 by driving a scanning system and an exposure system (neither are shown) for guiding image light from an original to an image sensing unit 1. Scanning and exposure arrangements and control are known to those who are skilled in the art, and a detailed description thereof will be omitted.

An analog image signal obtained by photoelectrically converting image light from an original by the image sensing unit 1 including an image sensing element such as a CCD is quantized into multi-value data by an A/D converter unit 2. Note that the A/D converter unit 2 outputs white level as a value "0", and outputs black level as a value "255". In the prescanning mode, a switch 6 is set at an adder 7 side under the control of the CPU 11.

The multi-value quantized data is input to the adder 7 to count frequencies of appearance in units of densities (to be referred to as a density histogram hereinafter). The count value is stored in a histogram memory 8. FIGS. 20A and 22A show contents of the histogram memory 8. FIG. 20A shows a density histogram of an image which entirely has a low density and a low contrast. FIG. 22A shows a density histogram of an image which entirely has a high density and a low contrast.

Upon completion of the prescanning operation, the CPU 11 looks up the content of the histogram memory 8, and outputs a correction table to a .gamma. correction table 22. FIGS. 20B and 20C and FIGS. 22B and 22C show this state.

More specifically, as shown in FIGS. 20A and 22A, if minimum and maximum densities appearing in an original are respectively represented by D.sub.min and D.sub.max, the CPU 11 looks up the histogram memory 8 to detect these D.sub.min and D.sub.max, and calculates a conversion table for performing the following three conversions according to the relationship among a value of output data D from the A/D converter unit 2, D.sub.min, and D.sub.max.

The CPU 11 calculates a .gamma. conversion table for D-D' conversion on the basis of equations [2] (see FIG. 20B or 22B).

A function F(D) is a conventional .gamma. correction function shown in FIG. 20C or 22C. In this embodiment, this function is compressed to fall within a range of densities present in an original, as shown in FIG. 20B or 22B.

In this manner, the calculation result is written in the .gamma. conversion table 22.

The CPU 11 switches the switch 6 to a .gamma. correction unit 21 side to perform a scanning operation for outputting an image signal. In this case, the multi-value quantized data D from the A/D converter unit 2 is input to the .gamma. correction unit 21, and is then .gamma.-corrected according to the .gamma. correction table 22. The .gamma.-corrected data is then subjected to tone processing such as dither processing by a tone processing unit 5, and the processed data is then output.

As a result, an image having a density histogram shown in FIG. 20A is tone-converted according to the .gamma. correction table shown in FIG. 20B, or an image having a density histogram shown in FIG. 22A is tone-converted according to the .gamma. correction table shown in FIG. 22B. Therefore, each of these image data is output as data D' as if it originally had a high-contrast density histogram, as shown in FIG. 21 or 23. More specifically, a reproduced image having a high contrast and is properly density-converted can always be obtained from a low-contrast original image.

Fifth Embodiment

In this embodiment, the CPU 11 detects D.sub.max and D.sub.min of a density whose frequency of appearance exceeds a predetermined value (N) from a density histogram generated by the prescanning operation like in the fourth embodiment, calculates a correction table according to equations [2] using the detected densities, and outputs the calculated correction table to the .gamma. correction table 22. FIGS. 24A and 24B show this state.

Thereafter, like in the fourth embodiment, level conversion is performed according to the correction data stored in the .gamma. correction table 22, and tone processing such as dither processing is performed by the tone processing unit 5.

In this embodiment, densities whose frequencies of appearance are below the predetermined value (N) are ignored in order to avoid level conversion errors caused by dust or scratches on an original. As the predetermined value N), an appropriate amount is used according to reading resolution of the image sensing unit 1 (e.g., a data volume per mm.sup.2), and a minimum size to be read as a highest (or lowest) density region. This embodiment is effective to prevent level conversion from being influenced by dust or scratches when an enlarged projection image of, e.g., a microfilm is to be read.

Sixth Embodiment

In this embodiment, when a density histogram is generated by the prescanning operation like in the fourth embodiment, a photometric region 26 is set at the central portion, and frequencies of appearance are counted in this region instead of counting frequencies of appearance of an entire reading region 23, as shown in FIG. 25.

A means for attaining this operation is realized by regulating a driving portion of a sensor such as a CCD included in the image sensing unit during the prescanning operation or by limiting a data input period to the adder 7 under the control of the CPU 11.

The CPU 11 looks up a density histogram generated for the central photometric region 26, and detects maximum and minimum densities D.sub.max and D.sub.min (or D.sub.max and D.sub.min for a density whose frequency of appearance exceeds a predetermined value as in the fifth embodiment). The CPU 11 then outputs correction data based on equations [2] using the detected D.sub.max and D.sub.min to the .gamma. correction table 22 like in the fourth embodiment.

Thereafter, like in the fourth embodiment, level conversion is performed according to the .gamma. correction table 22, and tone processing such as dither processing is performed by the gone processing unit 5.

In this embodiment, when an original has a small size and is present not in the entire image to be read by the image sensing unit 1, as shown in FIG. 25 (in FIG. 25, 24 denotes an original), a level conversion error caused by a portion 25 having no original in the image to be read can be avoided.

As described above, according to the fourth to sixth embodiments, Since the way of tone correction is changed according to frequencies of appearance in units of density levels of an original, a high-contrast tone image can be obtained from a low-contrast original without impairing an original tone. Since tone correction is performed using a conventional .gamma. correction mechanism, a mechanism to be added can be minimized.

Seventh Embodiment

FIG. 26 shows the seventh embodiment of the present invention.

The same reference numerals in FIG. 26 denote the same parts as in FIG. 9. An adder 7 serves as an appearance frequency detection means, and detects a frequency of appearance of each density level (to be referred to as a density histogram hereinafter). A histogram memory 8 stores a density histogram obtained by the adder 7. A CPU 11 determines dither patterns used in halftone processing of an image on the basis of the histogram stored in the histogram memory 8. A dither pattern memory 32 stores determined dither patterns. A tone processing unit 31 performs tone processing to obtain all the predetermined density levels on the basis of the dither patterns stored in the dither pattern memory 32. A switch 6 switches data from an A/D converter unit 2 to the adder 7 side in the prescanning mode, and switches it to the tone processing unit 31 side in the scanning mode.

An image reading apparatus of this embodiment performs a scanning operation for detecting a frequency of appearance of each density level (to be referred to as a prescanning operation) prior to a scanning operation for outputting image data.

(1) A control sequence executed by the CPU 11 when an image which entirely has a low density and a low contrast, as shown in FIG. 39, is to be reproduced will be described below.

In this case, a dither pattern having 4.times.4=16 tone levels is used.

A density histogram obtained by the prescanning operation is as shown in FIG. 27, and is stored in the histogram memory 8.

Upon completion of the prescanning operation, the dither patterns having characteristics shown in FIG. 28 are stored in the dither pattern memory 32 with reference to the content of the histogram memory 8.

More specifically, minimum and maximum densities D.sub.min and D.sub.max of an original are detected with reference to the histogram memory 8. Dither patterns for converting density data into dot numbers according to three conditions shown in Table 1 below in correspondence with the relationship among density data D, D.sub.max and D.sub.min are written in the dither pattern memory 32.

The CPU 11 then switches the switch 6 from the adder 7 side to the tone processing unit 31 side, and starts a scanning operation for outputting an image signal. The multi-value quantized data from the A/D converter unit 2 is input to the tone processing unit 31. The tone processing unit 31 converts density data into dot numbers according to the dither patterns stored in the dither pattern memory 32, so that the dot numbers have a density histogram shown in FIG. 29, and outputs an image shown in FIG. 33.

(2) A control sequence executed by the CPU 11 when an image which entirely has a high density and a low contrast, as shown in FIG. 40, is to be reproduced will be described below.

In comparison with the case (1) described above, a dither pattern to De stored in the dither pattern memory 32 is different. More specifically, a dither pattern having characteristics shown in FIG. 28 is used in the case (1), while a dither pattern having characteristics shown in FIG. 31 is used in this case. The operation and effect in this case are essentially the same as those in the case (1). In this case, an image shown in FIG. 33 can be output.

Eighth Embodiment

In the seventh embodiment, the CPU 11 rewrites the content of the dither pattern memory 32 with reference to the histogram memory 8. Alternatively, a .gamma. conversion unit 33 having a .gamma. conversion table memory whose I/O relationship can be rewritten by the CPU 11 is arranged between the switch 6 and the tone processing unit 31, as shown in FIG. 34, so that the content of the dither pattern memory 32 is fixed, and the content of the .gamma. conversion table memory may be rewritten to have an I/O relationship shown in FIG. 35.

The tone processing unit 31 need not always employ a dither method, but may employ other tone expression methods, e.g., a known error diffusion method.

Ninth Embodiment

In this embodiment, upon comparison with the seventh embodiment, a dither pattern to be stored in the dither pattern memory 32 is different. More specifically, in the seventh embodiment, a maximum density of an original corresponds to 16 dots as a maximum density D.sub.max, and a minimum density corresponds to 0 dot as a minimum density D.sub.min. However, in this embodiment, as shown in FIGS. 36 and 37, a maximum density of densities which are generated a predetermined number of times (N) or more corresponds to 16 dots as the maximum density D.sub.max, and a minimum density of densities which are generated the predetermined number of times (N) or more corresponds to 0 dot as the minimum density D.sub.min. As the value N, an appropriate amount is used according to a reading resolution of the image sensing unit 1 (e.g., a data volume per mm.sup.2), and a minimum size to be read as a highest (or lowest) density region. In this manner, the operation and effect of this embodiment are the same as those in the first embodiment. In addition, this embodiment is advantageous to prevent level conversion from being influenced by dust or scratches when an enlarged projection image of, e.g., a microfilm is to be read.

Tenth Embodiment

In this embodiment, upon comparison with the seventh embodiment, a region on an original for detecting a frequency of appearance of each density level is different. More specifically, in the seventh embodiment, such a detection region corresponds to the entire original shown in FIG. 38. In this embodiment, however, such a detection region corresponds to a central photometric region 42. In this manner, the operation and effect of this embodiment are the same as those in the seventh embodiment. In addition, when an original 44 is present not in the entire image to be read, appropriate tone conversion can be performed.

As described above, according to the seventh to tenth embodiments, a high-contrast tone image can be obtained from a low-contrast original without impairing an original tone.

Some preferred embodiments of the present invention have been described. However, the present invention is not limited to these embodiments, and various other changes and modifications may be made within the scope of claims.