JPH10117278A - Image reader - Google Patents

Abstract

PROBLEM TO BE SOLVED: To attain high gradation by devising the reader to process image data in bit number, providing a higher degree than a current 8-bit gradation number without increasing the cost much. SOLUTION: The reader is provided with a black reference data arithmetic means that obtains high degree 10-bit black reference data from an 8-bit A/D converter 21 A/D-converting data from a plurality of black reference photoelectric conversion elements provided in an image sensor, a black reference data memory 23 that stores the 10-bit black reference data obtained from the black reference data arithmetic means, an image data integral multiple means 24 that applies a four times image data to the image data outputted from the image sensor and A/D-converted to match the bit number with that of the black reference data, an arithmetic means 25 that subtracts the 10-bit black reference data from the 10-bit image data, and a black level correction circuit 22 that provides an output of image data which has been subjected to black level correction.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a digital copying machine,
The present invention relates to an image reading apparatus such as a facsimile apparatus that reads an original image using an image sensor.

[0002]

2. Description of the Related Art In general, an apparatus for photoelectrically converting and reading reflected light from a document by an image sensor such as a CCD is used.
Mainly, the illuminance distribution characteristics of the light source (that is, the distribution of the illuminance is higher at the center than at both ends of the light source) Condensing characteristics of the lens (the so-called cosine fourth law that the light is collected at the center of the lens) Sensitivity correction called shading correction is performed for reasons such as uneven sensitivity of each photoelectric conversion element (pixel) of the image sensor. For this shading correction, a reference white plate which is generally set outside the document image area and readable by an image sensor is used.

Conversely, a signal level that can be output from each photoelectric conversion element when the image sensor is not in a reading state is used as black reference data, and black correction is performed to match the level.

[0004] In recent years, in digital copiers and the like, there has been an increasing demand for higher image quality while maintaining the price of the main body. Speaking of high image quality is, for example, high resolution, and there is a plan to change the resolution from 400 DPI to 600 DPI. On the other hand, there is a trend toward higher gradation as high image quality, and there is a plan to increase the order from 8 bits to 10 bits.

[0005]

Such a plan is, of course, satisfactorily feasible with the current technology, but if it is used as it is, the cost will be considerably high. Taking high gradation as an example, in a scanner system, a 10-bit A / D
A converter, a shading correction operation circuit, and a memory therefor are required. The image processing system requires a 10-bit γ correction table, an inter-line correction memory, a separation operation circuit, a dither operation circuit, a magnification correction circuit, and the like. That is, the arithmetic circuit has two bits more than the current state, and the memory space is four times the current state. Also, these developments and designs are almost redones and require considerable manpower and time to implement.

As a practical problem, in a certain digital copying machine, there is a problem that a vertical streak is generated in a black portion in a full-color photographic mode, particularly at the time of enlargement. According to the analysis of the present inventor, it has been found that the cause is that a calculation error of 1 digit due to the shading correction calculation is expanded to 4 digits by the scanner γ.
Here, the “scanner γ” is for converting the data read by the scanner from the reflectance to the image density, and is usually written in the ROM in a table format. FIG.
Shows an example of the characteristics of the 8-bit scanner γ. FIG. 7 shows an example of an 8-bit configuration of a processing circuit in an image reading apparatus including a scanner γ conversion function according to such characteristics.
Schematically, an A / D converter 1, a black correction circuit 2, a shading correction circuit 3, and a scanner γ conversion circuit 4 are provided. The A / D converter 1 converts analog data input from an image scanner (not shown) into digital data and has an 8-bit resolution. The shading correction circuit 3 includes a shading correction data storage RAM 4.
, A ROM 5 and a multiplier 6. First, prior to reading a document, the reference white board read data (white reference data) that has been processed by the A / D converter 1 and the black correction circuit 2 by reading the reference white board by the image sensor is stored in the shading correction data storage RAM 4 pixel by pixel. Store it. Next, at the time of reading the original, the image sensor reads the reference white plate and processes the A / D converter 1 and the black correction circuit 2 to obtain black-corrected image data and shading correction data corresponding to pixels of the image data based on the RAM 4. Read from the ROM 5 and in the multiplier 6 image data after shading correction = read image data × 2
A multiplication process of 55 / white data is performed, and image data after shading correction is output. The image data after the shading correction is supplied to the scanner γ conversion circuit 4, and the image data after the scanner γ conversion that has been subjected to the scanner γ conversion processing is output to the signal processing system at the subsequent stage. Originally, the white reference data is read by, for example, 16 lines, and averaged by 8 lines to reduce the influence of dust on the reference white plate.
Bit white reference data is obtained.

FIG. 8 shows an example of such a shading correction result. The illustrated example shows the value of the correction result when the uniform black image data 5 is input when the shading correction data changes from 200 to 220, for example. Here, the shading correction data is 2
00 to 220 are the change widths of the read data of the reference white plate in a short cycle of about 10 pixels actually observed by a certain digital color copying machine, and the black image data 5 is the slope of the corresponding scanner γ. Is the value selected as the representative value of the largest data. According to FIG. 8, the shading correction data is divided between 211 and 212.
It can be seen that the correction result has changed from 5 to 6. In the shading correction stage, there is an operation error of 1 digit, but after the scanner γ conversion, there is a portion where conversion data has a large jump as shown as an enlarged portion A in FIG. 4 digit difference (error)
And the above-mentioned vertical streaks become conspicuous in this portion.

Accordingly, the present invention provides an image reading apparatus capable of achieving higher gradation by making it possible to handle image data of a higher-order bit number than the current 8-bit gradation number without increasing the cost so much. It is a first object to provide a device.

[0009] In addition, without significantly increasing the cost,
It is a second object of the present invention to provide an image reading apparatus capable of reducing the calculation error of the shading correction by obtaining the image data of the higher-order bit number and the white reference data and then performing the shading correction.

Further, it is possible to suppress the gradation jump when an input / output data conversion table such as a scanner γ conversion is provided, and to have no influence on the image processing after the input / output data conversion. It is a third object of the present invention to provide an image reading apparatus capable of reducing human and time waste in development relating to high gradation.

[0011]

According to a first aspect of the present invention, there is provided an image reading apparatus comprising: an image sensor having a plurality of photoelectric conversion elements;
An image reading apparatus provided with an A / D converter for performing D-conversion. The image sensor has a plurality of black reference photoelectric conversion elements in the image sensor, and outputs the A / D signals from these black reference photoelectric conversion elements. Black reference data calculation means for calculating data A / D converted by the D converter to obtain black reference data having a bit number higher than the resolution of the A / D converter; and black reference data calculation means for obtaining the black reference data calculation means. A black reference data memory for storing the obtained black reference data, and an image data output from the image sensor and A / D converted by the A / D converter by an integer multiple to match the number of bits of the black reference data. An image data integer multiplying means; and an arithmetic means for subtracting the black reference data stored in the black reference data memory from the image data output from the image data integer multiplying means. With a black correction circuit for outputting image data.

Therefore, the black reference data calculation means calculates data output from the plurality of photoelectric conversion elements for black reference in the image sensor and A / D converted by the A / D converter to perform the A / D conversion. Black reference data with a higher-order bit number than the resolution of the detector is obtained, but since the black reference data is originally a small value, the output data of a plurality of black reference photoelectric conversion elements is used. This does not necessarily require a large capacity, but only requires the same number of bits per pixel as the current state, so that the number of gray levels for black correction can be increased at the current cost.

According to a second aspect of the present invention, there is provided an image reading apparatus comprising: an image sensor having a plurality of photoelectric conversion elements; and an A / D converting an A / D signal output from the image sensor.
In an image reading apparatus provided with a converter, a plurality of times of data output from the photoelectric conversion elements for all pixels in the image sensor and A / D converted by the A / D converter are calculated in an unlit state. Black reference data calculating means for obtaining, for each pixel, black reference data having a bit number higher than the resolution of the leverage A / D converter; and black reference data for each pixel obtained by the black reference data calculating means. A black reference data memory for storing, and the A / D output from the image sensor.
Image data integer multiplying means for multiplying the image data A / D-converted by the converter by an integer to match the number of bits of the black reference data, and the black reference data memory from the image data output from the image data integer multiplying means And a calculating means for subtracting the black reference data of the corresponding pixel stored in the memory device, and outputting a black-corrected image data.

Accordingly, the black reference data calculation means calculates a plurality of times of data output from the photoelectric conversion elements for all the pixels in the image sensor and A / D converted by the A / D converter in a light-off state. Although black reference data of a higher order bit number than the resolution of the A / D converter is obtained for each pixel, since the black reference data is originally a small value, output data for a plurality of lines of photoelectric conversion elements for all pixels is obtained. Is not necessary to use a large capacity of the black reference data memory.
Since it is sufficient that the number of bits per pixel is the same as the current number, the number of gray levels for black correction can be increased at the current cost.

According to a third aspect of the present invention, in the image reading device according to the first or second aspect, a reference white plate for shading correction readable by an image sensor is provided, and the reference white plate is read to read the reference white plate. A plurality of times of data output from the photoelectric conversion elements for all pixels and A / D converted by the A / D converter are calculated, and white reference data having a bit number higher than the resolution of the A / D converter is calculated. A shading correction circuit is provided which has white reference data calculating means for obtaining each pixel and outputs image data obtained by shading correction of the image data black-corrected by the black correction circuit based on the white reference data of the corresponding pixel.

Therefore, similarly to the case of the black reference data, the shading correction is performed after obtaining the image data of a higher bit number and the white reference data at the current cost, so that the calculation error of the shading correction is reduced. Can be.

According to a fourth aspect of the present invention, in the image reading apparatus of the third aspect, the image data output by the shading correction circuit is converted into a signal of the A / D converter based on the number of higher-order bits in a stage subsequent to the shading correction circuit. A conversion processing unit having an input / output data conversion table set to convert to the same number of bits as the resolution is provided.

Therefore, the input / output data conversion table in the conversion processing section is set so as to convert the image data output by the shading correction circuit from a higher-order bit number to the same bit number as the resolution of the A / D converter. Therefore, skipping of gradation can be suppressed, and the output has the same number of bits as the resolution of the A / D converter. Human and time wasted in development related to industrialization.

[0019]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described with reference to FIGS. First, the mechanical configuration of the image reading apparatus according to the present embodiment and its basic operation will be described with reference to FIG. In the image reading apparatus according to the present embodiment, an original (not shown) set on a contact glass 11 is irradiated with illumination light from a light source 12 such as a halogen lamp, and the reflected light is applied to first to third light sources. Mirrors 13, 14, 1
An image is irradiated on the light receiving surface of the image sensor 17 through the imaging lens 5 and the imaging lens 16. Here, the light source 12 and the first mirror 13 are mounted on a first carriage 18, and the second and third mirrors 14 and 15 are mounted on a second carriage 19, and perform sub-scanning in the same direction at a speed ratio of 2: 1. Is set to The image sensor 17 is, for example, a CCD line sensor, and a large number of photoelectric conversion elements are linearly arranged on a light receiving surface thereof. Further, a reference white plate 20 which is located outside the contact glass 11 and can be read by the image sensor 17 is provided.

Next, an example of a circuit configuration of an image processing system for processing an analog signal output from the image sensor 17 is shown in FIG.
This will be described below. In the present embodiment, it is assumed that the current 8-bit configuration is upgraded to a 10-bit configuration.
First, in the image sensor 17, a photoelectric conversion element for a part of four pixels is a photoelectric conversion element for black reference, and has a light shield structure that cannot receive external light.
An A / D converter 21 is connected to the output side of the image sensor 17. The A / D converter 21 converts an analog signal output from the image sensor 17 into digital data and has the same 8-bit resolution as the current state. This A / D converter 2
1 is connected to a black correction circuit 22. The black correction circuit 22 includes a black reference data calculation means (not shown) and a black correction data storage RAM 2 serving as a black reference data memory.
3 and a quadruple circuit 24 as an image data integer multiplying means and a subtraction circuit 25 as an arithmetic means. The black reference data calculation means performs a calculation for adding and averaging the data output from the four light-shielded photoelectric conversion elements for black reference and A / D converted by the A / D converter 21. It functions to obtain the number of bits higher than the resolution (8 bits) of the A / D converter 21, here, 10-bit black reference data. The black correction data storage RAM 23 stores the 10-bit black reference data. However, since the black reference data is originally small, the target memory for four pixels does not mean that the required memory capacity is increased, but one byte (= 8 bits) per pixel is sufficient. May be added with “00” as the upper two bits. On the other hand, the quadruple circuit 24 quadruples the 8-bit image data read by the image sensor 17 and A / D-converted by the A / D converter 21 to obtain 10-bit image data. The subtraction circuit 25 performs an arithmetic process of subtracting the black reference data stored in the black correction data storage RAM 23 from the image data of each pixel which has been made 10 bits by the quadruple circuit 24.

The output side of the black correction circuit 22 is connected to a shading correction operation circuit 26 which is a shading correction circuit. The shading correction operation circuit 26 includes a shading correction data storage RAM 27, a ROM 28, and a multiplier 29 as in the case shown in FIG.
The multiplier 28 and the multiplier 29 have a 10-bit specification. “INT” in the ROM 28 and the multiplier 29 means an integer conversion process in which a fractional part below the decimal point is ignored. A scanner γ conversion circuit 30 serving as a conversion processing unit is connected to an output side of the shading correction operation circuit 26. The scanner γ-conversion circuit 30 is configured to convert the shading-corrected image data output from the shading-correction operation circuit 26 from 10 bits to 8 bits, the same as the resolution of the A / D converter 21, and to set input / output data A table for the scanner γ, which is a conversion table, is provided.

In such a configuration, prior to reading a document, the reference white board 20 is read by the image scanner 17 and the reference white board read data (white reference data) that has been processed by the A / D converter 21 and the black correction circuit 22 is output as one. Each pixel is stored in the shading correction data storage RAM 27 as shading correction data. Also, the black correction circuit 22
In the black correction data storage RAM 23, digital data obtained by A / D conversion of the signal from the light-shielded black reference photoelectric conversion element by the A / D converter 21 is used as black reference data.
The black correction is performed by subtracting the value of the black reference data from the image data of each pixel by the subtraction circuit 25 when reading the original image. This method of black correction is called, for example, line black correction.

Here, with respect to such basic line black correction, in the present embodiment, the current 8 bits are increased to 10 bits and are processed as follows. First,
The outputs of the light-shielded photoelectric conversion elements for black reference for four pixels are A / D converted by the A / D converter 21 and provided to the black correction circuit 22 as 8-bit digital data. In the black correction circuit 22, 10-bit black reference data is obtained by adding and averaging the outputs of the four pixels of the black reference photoelectric conversion elements by the black reference data calculation means. It is stored in the storage RAM 23. In this case, since the black reference data is originally a small value, the use of the output data of the four black reference photoelectric conversion elements does not necessarily require a large capacity of the black correction data storage RAM 23, but one. Since it is sufficient that the number of bits per pixel is the same as the current 8 bits, the number of gray levels for black correction can be increased at the current cost. That is, since the original image data is quadrupled by the quadrupling circuit 24 to make 10 bits, the lower 2 bits are invalid, but the 10-bit black reference data is converted from such 10-bit image data. Since the subtraction is performed in the subtraction circuit 25, it can be said that 10 bits of the black-corrected image data after the subtraction are effective to some extent.

This is executed by the image data black correction unit 31 functionally shown in the black correction circuit 22 in FIG.

Next, a description will be given of the suppression of calculation errors in shading correction. Regarding the shading correction, when reading and averaging the reference white plate 21 over a plurality of lines, the white reference data calculation means converts the shading correction data from 8 bits to 10 bits to higher order bits. That is, in order to reduce the influence of dust or the like on the reference white plate 20, the reference white plate 20 is read for 16 lines and averaged to obtain the current 8 bits. This is averaged using the ROM 28. In this case, data of up to 10 bits may be taken. That is, it is a matter of how many bits the result of the addition takes, and the capacity of the RAM 27 for storing shading correction data does not need to be large. This corresponds to the black correction circuit 22 in FIG.
The processing is executed by the shading correction data black correction unit 32 shown functionally therein. The 10-bit shading correction data thus obtained is supplied to the A / D converter 2
1 can be said to be valid data that is not different from that obtained by A / D conversion with the 10-bit specification.

FIG. 4 shows an example of the result of the shading correction calculation, as in the case described with reference to FIG. In the illustrated example, the shading correction data and the black image data are 8
This is four times the bit size, and the shading correction data is, for example, 800 (= 200 × 4) to 880 (= 220 ×
4) When the change width is as follows, 20 as the black image data
It shows the value of the correction result when taking (= 5 × 4).
However, in the case of 10 bits, since the black image data is further subdivided, it is not uniform. For example, when the shading correction data is 800, a value proportional to the reference value is 20, for example, when the shading correction data is 840, 21,880
In the case of, it is set to 22 and so on. Originally, the read data of the reference white board 20 and the original image data are proportional, and the shading correction is used to correct the unevenness of the sensitivity of the pixels and the unevenness of the illuminance distribution of the illumination by dividing the image data by the reference white board data. It is a purpose and is in line with this purpose. As a result, in the case of 8-bit processing, as described with reference to FIG. 8, even if the shading correction data changes from 200 to 220, the corresponding black image data does not change at 5, but the 10-bit processing according to the present embodiment. In this case, it can be seen that the correction result repeats 25 and 24 in a certain cycle. Here, the calculation error of the shading correction is 1 digit of 5 and 6 in the case of 8 bits, and 1 digit of 24 and 25 in the case of 10 bits. Then, it can be seen from the comparison between FIG. 4 and FIG. Therefore, according to the present embodiment, since the shading correction calculation by the multiplier 29 is performed after obtaining the 10-bit image data and the shading correction data, the calculation error of the shading correction is suppressed. Is suppressed to 1/4.

Next, the scanner γ conversion circuit 30 will be described. The scanner γ conversion circuit 30 receives the 10-bit image data after shading correction from the shading correction operation circuit 26 and performs an input / output conversion process for converting the reflectance into a density. Since the conversion processing is performed with the contents that become bit outputs, the image processing after the scanner γ conversion circuit 30 is not affected at all, and the conventional configuration is sufficient. That is, the former stage of the black correction circuit 22 and the latter stage of the scanner γ conversion circuit 30 have the conventional 8-bit specification.

FIG. 5 shows the scanner γ characteristics when a table of 10 bits input and 8 bits output is used as the scanner γ. According to FIG. 5, it can be seen that the gradation skip seen in the vicinity of the portion A in FIG. 6 is eliminated.
This is because, since the number of input data is quadrupled, the inclination of 4 digits at maximum as in FIG. 6 can be suppressed to 1 digit. Therefore, vertical streaks that may occur in a black portion due to a calculation error of shading correction appear only with a density difference of 現状 of the current level, are almost inconspicuous as compared with the current level, and high-quality high gradation can be achieved.

In this embodiment, the black correction is a line black correction method, and a large number of photoelectric conversion elements of the image sensor 17 include a light shielding black reference photoelectric conversion element. Alternatively, black correction may be performed by a pixel black correction method. That is, with the light source 12 turned off, the data output from the photoelectric conversion elements for all the pixels of the image sensor 17 and A / D converted by the A / D converter 21 are used as black reference data for black correction for each pixel. In this method, the image data is stored in the data storage RAM 23, and the image data subjected to black correction is output by subtracting the black reference data of the corresponding pixel from the image data when reading the image.
In the case of this pixel black correction method, in order to convert the black reference data into 10 bits, the operation of obtaining outputs from the photoelectric conversion elements for all the pixels of the image sensor 17 with the light source 12 turned off is required.
By repeating the data four times and averaging the data, black reference data of 10 bits for each pixel may be acquired and stored in the black correction data storage RAM 23.
Other processes are the same as those in the case of the line black correction method.

[0030]

According to the first aspect of the present invention, the image sensor has a plurality of black reference photoelectric conversion elements, and the output from these black reference photoelectric conversion elements is output by the A / D converter. Black reference data calculation means for calculating the A / D converted data to obtain black reference data having a bit number higher than the resolution of the A / D converter; and a black reference data obtained by the black reference data calculation means. A black reference data memory for storing data, image data integer multiplying means for multiplying the image data output from the image sensor and A / D converted by the A / D converter by an integer to match the number of bits of the black reference data, Calculating means for subtracting the black reference data stored in the black reference data memory from the image data output from the image data integer multiplying means, and a black correction circuit for outputting black-corrected image data. According to the second aspect of the present invention, in a non-lighting state, a plurality of times of data output from the photoelectric conversion elements for all pixels in the image sensor and A / D converted by the A / D converter are calculated. Black reference data calculation means for obtaining, for each pixel, black reference data having a higher bit number than the resolution of the A / D converter, and black reference data for each pixel obtained by the black reference data calculation means are stored. A black reference data memory; image data integer multiplying means for multiplying the image data output from the image sensor and A / D converted by the A / D converter by an integer to match the number of bits of the black reference data;
Computing means for subtracting the black reference data of the corresponding pixel stored in the black reference data memory from the image data output from the image data integer multiplying means, and outputting black-corrected image data. Since the circuit is provided, it is possible to obtain black-corrected image data of a higher-order bit number than the current number of gradations without increasing the cost so much that higher gradations can be achieved.

According to the third aspect of the present invention, in addition to the image reading device of the first or second aspect, a reference white plate for shading correction readable by an image sensor is provided. A plurality of data output from the photoelectric conversion elements for all pixels in the sensor and A / D-converted by the A / D converter are calculated, and white data of a higher-order bit number than the resolution of the A / D converter is calculated. A shading correction circuit having white reference data calculating means for obtaining reference data for each pixel, and outputting image data obtained by shading correction of the image data black-corrected by the black correction circuit based on the white reference data of the corresponding pixel; As in the case of black reference data for black correction, image data of a higher-order bit number and white reference data are obtained at the current cost, and then shaded. Since graying correction, it is possible to reduce the calculation error of the shading correction.

According to a fourth aspect of the present invention, in the image reading apparatus according to the third aspect, the image data output by the shading correction circuit is A / D-converted from a higher-order bit number in a subsequent stage of the shading correction circuit. A conversion processing unit having an input / output data conversion table set so as to convert to the same number of bits as the resolution of the device, it is possible to suppress the skip of gradation and output the A / D converter. Since the number of bits is the same as the resolution, the image processing after the conversion processing unit does not differ from the conventional one, and human and time waste can be reduced in development relating to higher gradation.

[Brief description of the drawings]

FIG. 1 is a block diagram of an image processing unit according to an embodiment of the present invention.

FIG. 2 is a schematic front view illustrating a mechanical configuration of the image reading apparatus.

FIG. 3 is a block diagram partially showing a functional configuration;

FIG. 4 is a characteristic diagram illustrating a calculation error of shading correction in the case of 10 bits.

FIG. 5 is a characteristic diagram showing a scanner γ in a case of 10-bit input and 8-bit output.

FIG. 6 is a characteristic diagram showing a scanner γ in the case of a conventional 8-bit input / 8-bit output.

FIG. 7 is a block diagram illustrating a configuration of a conventional image processing unit.

FIG. 8 is a characteristic diagram showing a calculation error of shading correction in the case of a conventional 8-bit.

[Explanation of symbols]

 Reference Signs List 17 image sensor 20 reference white plate 21 A / D converter 22 black correction circuit 23 black reference data memory 24 image data integral multiple circuit 25 arithmetic means 26 shading correction circuit 30 conversion processing unit

Claims (4)

[Claims] An image sensor having a large number of photoelectric conversion elements, and a signal output from the image sensor is A / A
An image reading apparatus comprising: an A / D converter for performing D-conversion; a plurality of black-reference photoelectric conversion elements in the image sensor;
Black reference data calculation means for calculating data A / D converted by the D converter to obtain black reference data having a bit number higher than the resolution of the A / D converter; and black reference data calculation means for obtaining the black reference data calculation means. A black reference data memory for storing the obtained black reference data, and an image data output from the image sensor and A / D converted by the A / D converter by an integer multiple to match the number of bits of the black reference data. Image data integer multiplying means, and arithmetic means for subtracting the black reference data stored in the black reference data memory from the image data output from the image data integer multiplying means. An image reading device comprising a black correction circuit for outputting. 2. An image sensor having a large number of photoelectric conversion elements, and a signal output from the image sensor is A / A
An image reading apparatus having an A / D converter for performing D-conversion, wherein a plurality of signals output from photoelectric conversion elements for all pixels in the image sensor and A / D-converted by the A / D converter in an unlit state. Black reference data calculating means for calculating the data for each batch to obtain black reference data of a bit number higher than the resolution of the A / D converter for each pixel; and each pixel obtained by the black reference data calculating means. A black reference data memory for storing black reference data for each image, and image data output from the image sensor and A / D converted by the A / D converter by an integer multiple to match the number of bits of the black reference data. Image data integer multiplying means, and arithmetic means for subtracting black reference data of a corresponding pixel stored in the black reference data memory from image data output from the image data integer multiplying means. An image reading apparatus comprising: a black correction circuit that outputs black-corrected image data. 3. A reference white plate for shading correction which can be read by an image sensor. By reading the reference white plate, the signals are output from the photoelectric conversion elements for all pixels in the image sensor, and are output by an A / D converter. A black correction circuit for calculating a plurality of / D-converted data to obtain white reference data having a higher-order bit number than the resolution of the A / D converter for each pixel; 3. The image reading device according to claim 1, further comprising a shading correction circuit that outputs image data obtained by performing shading correction on the image data that has been black-corrected based on white reference data of a corresponding pixel. 4. An input / output data set at a subsequent stage of the shading correction circuit so as to convert image data output by the shading correction circuit from a higher-order bit number to a bit number equal to the resolution of an A / D converter. The image reading apparatus according to claim 3, further comprising a conversion processing unit having a conversion table.