JP4895391B2 - Image reading device - Google Patents

Description

  The present invention relates to an image reading apparatus that optically reads an original, and more particularly to an image reading apparatus that is characterized by a technique for correcting a linearity defect of an image signal output from a CMOS sensor.

  A CCD sensor is generally used as a photoelectric conversion device of an image scanner such as a copying machine.

  In recent years, CMOS sensors have been widely used for photoelectric conversion devices of digital still cameras. The CMOS sensor has an advantage that it operates at the same low voltage as that of the CMOS logic, and that a general-purpose process can be applied in manufacturing. However, since the CMOS sensor has a configuration in which an amplifier is provided for each photodiode, fixed pattern noise is likely to occur due to a characteristic difference between the amplifiers.

  When the CMOS sensor is arranged one-dimensionally and a CMOS line sensor having a function equivalent to that of a conventional CCD line sensor is mounted on an image scanner, the above-described difference in amplifier characteristics may occur as an image defect.

  FIG. 1 is a diagram schematically illustrating an example of an input / output characteristic difference between amplifiers in a CMOS sensor.

  When the characteristic difference of the amplifier has linearity, it can be corrected by white shading and black shading which are gain correction for each pixel which is generally performed in an image scanner. However, in the CMOS sensor, as shown in FIG. 6, there is a case where the relationship between the incident light quantity and the output signal level shows a non-linear characteristic.

  In such a case, if the white and black shading correction is performed, the black level and the white level can be adjusted for each pixel, but sufficient correction is not performed in the halftone. As a result, the output image data appears as an image sub-scanning streak at pixel positions having nonlinear characteristics.

Since variations in the characteristics of amplifiers lead to a reduction in image quality, various countermeasures have been proposed to improve the characteristics. Patent Document 1 proposes the following technique for detecting pixel defects. That is, the output level of each element is expressed as a linear function y = ax + b, and a pixel defect is found by comparing with the preset constants a0 and b0, and an optimal constant is set for the pixel to correct the image. It is what.
JP 2000-92397 A

  The above-described conventional technique detects an inclination with respect to a linear function y = ax + b serving as a reference and a deviation from an intercept, and is an effective technique in an area sensor of a digital still camera that does not perform shading correction.

  However, pixel defects to be detected can be corrected by black shading correction and white shading correction for each pixel performed by an image scanner having a sensor. What is a detection / correction method suitable for an image scanner? I can't say that.

  An object of the present invention is to provide an image reading apparatus that corrects image streaks due to poor linearity of an amplifier of a CMOS line sensor and realizes a good image.

To achieve the above object, an image reading apparatus according to claim 1 , comprising: a light source for illuminating a document; and a CMOS line sensor that converts light reflected from the document by the light source into an electrical signal. in the apparatus, a calculation means for calculating a white shading correction coefficient a white reference plate that is illuminated by the light source by reading by the CMOS line sensor, and changing means for changing the light quantity of the light source, the light source by said changing means After the light amount is changed, the white reference plate is read by the CMOS line sensor, and the shading correction is performed on the image data obtained by reading the original by the CMOS line sensor using the white shading correction coefficient calculated by the calculation unit. a correction unit for performing, after the shading correction by the correction means The linearity inspection means for inspecting the linearity of the image by comparing the image data with a predetermined value for each pixel, and the pixel position where the linearity inspection is detected by the linearity inspection means and the output level of the CMOS line sensor are stored. Storage means for calculating the correction coefficient from the input level of the CMOS line sensor and the output level of the CMOS line sensor stored in the storage means, and storing the correction coefficient in the storage means; and the CMOS line sensor Linearity correction means for reading out the position of the pixel and the correction coefficient corresponding to the position of the pixel from the storage means and performing correction processing on the read image data of the original by the CMOS line sensor when reading the original by , characterized in that it comprises a.

The image reading apparatus according to claim 2, wherein the image reading apparatus includes a light source for illuminating the document and a CMOS line sensor that converts light reflected from the document by the light source into an electric signal. A white shading correction coefficient is calculated by reading the white reference plate with the CMOS line sensor; a changing means for changing the storage time of the CMOS line sensor; and a storage time of the CMOS line sensor by the changing means. After the change, the white reference plate is read by the CMOS line sensor, and the shading correction is performed on the image data obtained by reading the original by the CMOS line sensor using the white shading correction coefficient calculated by the calculation means. a correction unit, Shede by the correcting means The image data after the ring corrected, and linearity inspection means for inspecting the linearity of the image data by comparing with a predetermined value for each pixel, the position and the CMOS line sensor pixels linearity defect is detected by the linearity test means storage means for storing the output level, the correction coefficient calculation means for storing in the storage means to calculate a correction factor from the output level of the CMOS line sensor stored in the input level and the storage means of the CMOS line sensor When the original is read by the CMOS line sensor , the position of the pixel and the correction coefficient corresponding to the position of the pixel are read from the storage means, and the correction processing is performed on the read image data of the original by the CMOS line sensor. And linearity correction means for performing To.

  According to the present invention, it is possible to correct an image streak caused by a poor linearity of an amplifier of a CMOS line sensor and realize a good image.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  A method for detecting a pixel having a poor linearity of the amplifier will be described. In order to determine whether or not the linearity is normal, a plurality of signal levels obtained by reading reference white or the like from a state in which the signal level output from the CMOS sensor is 0 and a plurality of intermediate signal levels are acquired. And it determines by whether those data have linearity.

  At the start of detection of defective linearity pixels, a shading correction coefficient is calculated. Specifically, the black shading coefficient is calculated from the CMOS sensor output level when the light source for illuminating the document is turned off or the CMOS sensor output when the black reference plate is read.

  Subsequently, the light source is transferred, the white reference is read, and the white shading coefficient is calculated. In the CMOS sensor output subjected to shading correction by the white and black shading coefficients, the black level and white level of all pixels are uniform by a predetermined value.

  Next, a CMOS sensor output that is intermediate between black and white is acquired. There are various methods for outputting a halftone to the CMOS sensor.

  One is a method of controlling the light quantity of the light source. When the light amount of the light source at the time of generating the white shading correction coefficient is 100%, for example, the output level of the CMOS sensor is acquired in a state where the light source is turned on with a light amount of 25%, 50%, and 75%.

  The other is a method for controlling the accumulation time. For example, the accumulation time setting such as 2 times, 3 times, 4 times, etc. is performed with respect to the accumulation time when reading a normal document (image), and the output level of the CMOS sensor at each accumulation time setting is acquired. is there. If the 4 times setting is set to 100%, if the accumulation time setting is set to the normal setting, 2 times, or 3 times, it is equivalent to reading with the above-mentioned light amount change 25%, 50%, 75%. It becomes processing.

  FIG. 2 is a diagram illustrating an output characteristic example of the CMOS linear sensor obtained by adjusting the light amount of the light source or changing the accumulation time.

2, white circles, and black circles light amount adjustment, or Ru Tei represents the output level of the CMOS sensor obtained when changing the accumulation time. The white circle is when the output characteristic of the CMOS sensor is normal, and the black circle is the output characteristic of a pixel with poor linearity.

  When processing is performed such that the white level after shading correction is 255 at 256 gradations, the CMOS sensor output after shading correction has the following input / output characteristics if the linearity of the amplifier is normal.

  Near the 255th level when the light intensity is set to 100% (or 4 times the accumulation time), near the 191st level when the light intensity is set to 75% (the accumulation time is set to 3 times), near the 127th level when the light intensity is set to 50% (the accumulation time is set twice) If the setting is 25% (accumulation time is set to 1 time), it is around 63 levels.

  On the other hand, since the linearity is deteriorated in a pixel having a poor linearity, the output level when the light amount is changed or the accumulation time is changed is a value that is greatly different from the level value described above.

  In the present invention, the data value that is output when the linearity of the amplifier is normal is used as a theoretical value, and it is determined how far the output data of each pixel is from the theoretical value. To discover. A linearity correction coefficient is calculated from the output level.

  FIG. 3 is a block diagram of the image reading apparatus according to the embodiment of the present invention.

  In FIG. 3, the image reading apparatus controls a light source (illuminating unit) 301 that illuminates a document, a lighting circuit 302 that drives and drives the light source 301, a timing generator (TG) 303, and a lighting circuit 502. A CPU 304 for controlling the amount of light is provided.

  The image reading apparatus also includes a CMOS line sensor (one-dimensional CMOS image sensor) 305 that converts reflected light from a document into an electrical signal, a shading correction unit 306, and a linearity detection unit 307. The image reading apparatus also includes a correction coefficient calculation unit 308, an EEPROM 309, and a linearity correction unit 310.

  Here, the CPU 304 as a control unit functions as a light amount control unit that controls the light amount of the light source 301 and as an accumulation time control unit that controls the accumulation time of the CMOS line sensor 305.

  The TG 303 is a circuit that generates a timing signal for driving the CMOS line sensor 305, and can generate various timing signals under the control of the CPU 304. Therefore, it is possible to generate a drive signal whose storage time is changed under the control of the CPU 304.

  The output signal of the CMOS line sensor 305 is input to the shading correction unit 306 after analog signal processing such as gain processing and offset processing, and digital signal processing such as AD conversion.

  The shading correction unit 306 performs a light shading state or black shading correction using a black shading coefficient generated by reading the black reference, and white shading processing using a white shading coefficient generated by turning on the light source and reading the white reference.

  This is a circuit that performs linearity correction on image data output from the linearity correction unit 310 and the shading correction unit 306.

  The EEPROM 309 functions as a holding unit that holds the pixel position of the image data based on the detection result of the linearity detection unit 307 and the image data level when the light amount is controlled by the CPU 304 as the light amount control unit. The EEPROM 309 functions as a storage unit that stores the main scanning position detected by the linearity detection unit 307 and the correction coefficient calculated by the correction coefficient calculation unit 308.

  FIG. 4 is a diagram showing an outline of processing of the linearity correction unit in FIG.

  When an instruction to start reading a document (image) is issued, the CPU 304 reads out the pixel position (n-th pixel) to be corrected from the EEPROM 309 and a correction coefficient corresponding to the pixel position, and sets them in the linearity correction unit 310.

  From the shading correction unit 306, together with the image data after shading correction, image counter data synchronized with the main scanning order of the image data is output and output to the linearity correction unit 310.

  The counter data is compared with the set value n of the pixel position to be corrected by the linearity correction unit 310, and when the counter data matches the set value n of the pixel position, correction processing is performed on the corresponding image data Dn. carry out.

  If the counter data and the set value of the pixel position do not match, the correction process is unnecessary, and the input image data is passed through. Note that it is possible to perform linearity defect detection and correction processing for a plurality of pixels with respect to pixels for one line of the CMOS line sensor 305.

  The correction process is performed according to the input / output characteristics shown in FIG. A one-dimensional LUT method corresponding to the bit depth of the image or a correction coefficient prepared for input levels 0, 64, 127, 191, 255 is used. The input image data between them is calculated by linear interpolation of correction coefficients corresponding to adjacent input image data.

  FIG. 5 is a flowchart showing the procedure of the detection process of linearity failure pixels (linearity detection / correction coefficient calculation process) executed by the image reading apparatus of FIG.

  This process is executed by the CPU 304 in FIG.

  When the detection processing start is instructed, shading correction coefficient calculation is performed (calculation means) in step S101. A black shading correction coefficient is calculated by performing a reading operation with the light source 301 turned off, and a white shading correction coefficient is calculated by reading the white reference plate with the light source 301 turned on. At this time, the light source light amount and the accumulation time are set so that the output of the CMOS line sensor 305 is maximized.

  In step S102, the light source light amount or the accumulation time is changed (changing means). When changing the light source light amount and the accumulation time, settings are made so as to be smaller than the sensor output level in step S101. For example, when the light source amount in step S101 or the sensor output at the accumulation time setting is 100%, the sensor output is set to 75%.

  In step S103, the white reference plate (white plate) is read. At this time, shading correction is performed using the white and black shading coefficients calculated in step S101 (correction means).

  In step S104, level determination is performed (determination means). If the light source light amount or the accumulation time setting in step S102 is set to 75%, the sensor output value after shading correction is a normal output value at 191 level which is 75% of 256 gradation 255 level. Level judgment is performed using the value as a reference level (reference value).

  The reference level may be set to have a certain range in consideration of variations in the reading level. The level determination is performed for all pixels of the CMOS line sensor 305, and determination is performed for each pixel. If there is a pixel that is outside the reference level as a result of the determination, an NG determination is made, and the process of step S104 is performed. .

  In step S105, a linearity failure is detected based on the level determination result in step S103 (detection means). And the process which memorize | stores the information of the pixel of NG is performed. The stored information is the pixel position and output level, and is used to calculate the correction coefficient and determine the correction position when performing the later linearity correction processing.

  In step S106, an execution determination process for the light source quantity or the accumulation time setting process is performed. If there is still a setting condition of the light source quantity to be determined or the accumulation time, the process proceeds to step S102, and the next condition setting is performed. For example, after the level determination processing at 75% setting, processing is performed so that 50% setting and 25% setting are sequentially performed.

  If the processing under all conditions has been completed, this flow ends.

  When the above flow is performed, the output of the shading correction unit 306 is sent to the linearity detection unit 307. The image data sent to the linearity detection unit 307 is compared with a predetermined value pixel by pixel, and the linearity of the image is inspected (detected).

  When it is determined that the linearity is broken, the determined pixel position is stored, and the image data is sent to the correction coefficient calculation unit 308, and the correction coefficient is calculated from the ratio between the image data and a predetermined value. Thereafter, the pixel position and the correction coefficient are stored in the EEPROM 309.

  As a result of performing the processing flow of FIG. 5, a pixel determined to be a poor linearity pixel has an upward convex characteristic with respect to the normal characteristic as shown in FIG. 2.

  FIG. 6 is a diagram showing correction coefficients for pixels having convex characteristics on the upper side in FIG.

  In FIG. 6, the horizontal axis indicates the image data level (input level) input to the linearity correction unit 310, the vertical axis indicates the correction coefficient, the solid line indicates the correction coefficient characteristic, and the input levels are 25%, 50%, and 75%. It becomes a polyline characteristic with an inflection point.

  The correction coefficient corresponding to the inflection point is calculated from the ratio between the normal output level and the output level acquired in the actual detection processing flow. For example, when the light amount setting (or accumulation time setting) is 25%, the normal input level is 63, whereas 116 is input.

  In this case, 63/116 = 0.543 is prepared as a correction coefficient in order to set the input level 116 to the normal input 63 level. Similarly, the correction coefficient is calculated under the light amount conditions (or accumulation time conditions) of 50% and 75%.

  Next, an input level under three conditions and a correction coefficient at an input level other than the 0 level and the 255 level are calculated from the correction coefficients calculated under the three conditions. Specifically, in a linear function approximation method, for example, a linear approximation expression is obtained from the difference between the level difference between the input level 63 with 25% light intensity (accumulation time) and the input level 127 with 50% setting and the correction coefficient. Generate.

  Then, correction coefficients from 64 levels to 126 are calculated from the approximate expression. The correction coefficient is calculated in the same manner for other input data ranges, but the correction coefficient is calculated assuming that the output level is 0 when the input data level is 0, and the output level is 255 when the input data level is 255. To do.

  As described above, in the present invention, the linearity defect of the amplifier of the CMOS line sensor 305 is detected for each pixel by changing the light amount condition of the light source 301 or the accumulation time condition of the CMOS line sensor 305 and reading the reference white plate. To detect. Since correction is performed for each pixel, it is possible to suppress deterioration in image quality such as sub-scanning streaks due to poor linearity.

  In addition, detection processing is possible with the image reading device alone, and no detection member such as a chart is required for detection of linearity failure. Therefore, even when the optical unit equipped with the CMOS line sensor 305 is replaced on the market, detection of linearity failure is possible. Correction is possible.

It is a figure which shows typically an example of the characteristic difference of the amplifier in a CMOS sensor as an input-output characteristic. It is a figure which shows the output characteristic example of a CMOS linear sensor obtained by adjusting the light quantity adjustment of a light source, or changing accumulation time. 1 is a block diagram of an image reading apparatus according to an embodiment of the present invention. It is a figure which shows the outline of a process of the linearity correction | amendment part in FIG. FIG. 4 is a flowchart showing a procedure of detection processing (linearity detection / correction coefficient calculation processing) of a linearity defect pixel executed by the image reading apparatus of FIG. 3. It is a figure which shows the correction coefficient with respect to the pixel of a convex characteristic on the upper side in FIG.

Explanation of symbols

301 Light source 302 Lighting circuit 303 Timing generator (TG)
304 CPU
305 CMOS line sensor 306 Shading correction unit 307 Linearity detection unit 308 Correction coefficient calculation unit 309 EEPROM
310 Linearity correction unit

Claims (2)

In an image reading apparatus comprising a light source for illuminating a document, and a CMOS line sensor that converts light reflected from the document by the light source into an electrical signal,
Calculating means for calculating a white shading correction coefficient by reading the white reference plate illuminated by the light source with the CMOS line sensor ;
Changing means for changing a light amount of the light source,
After the light quantity of the light source is changed by the changing means, the white reference plate is read by the CMOS line sensor, and the original is read by the CMOS line sensor using the white shading correction coefficient calculated by the calculating means. Correction means for performing shading correction on the obtained image data ;
Linearity inspection means for inspecting image linearity by comparing image data after shading correction by the correction means with a predetermined value for each pixel;
Storage means for storing the position of the pixel where the linearity defect is detected by the linearity inspection means and the output level of the CMOS line sensor;
Correction coefficient calculation means for calculating a correction coefficient from the input level of the CMOS line sensor and the output level of the CMOS line sensor stored in the storage means and storing the correction coefficient in the storage means;
When the original is read by the CMOS line sensor, the position of the pixel and the correction coefficient corresponding to the position of the pixel are read from the storage unit, and correction processing is performed on the read image data of the original by the CMOS line sensor. Linearity correction means;
An image reading apparatus comprising: In an image reading apparatus comprising a light source for illuminating a document, and a CMOS line sensor that converts light reflected from the document by the light source into an electrical signal,
Calculating means for calculating a white shading correction coefficient by reading the white reference plate illuminated by the light source with the CMOS line sensor;
Changing means for changing the accumulation time of the CMOS line sensor;
After the storage time of the CMOS line sensor is changed by the changing means, the white reference plate is read by the CMOS line sensor, and the original is read by the CMOS line sensor using the white shading correction coefficient calculated by the calculating means. Correction means for performing shading correction on image data obtained by reading; and
Linearity inspection means for inspecting the linearity of the image data by comparing the image data after shading correction by the correction means with a predetermined value for each pixel;
Storage means for storing the position of the pixel where the linearity defect is detected by the linearity inspection means and the output level of the CMOS line sensor;
A correction coefficient calculation means for storing in the storage means to calculate a correction factor from the output level of the CMOS line sensor stored in the input level and the storage means of the CMOS line sensor,
When the original is read by the CMOS line sensor , the position of the pixel and the correction coefficient corresponding to the position of the pixel are read from the storage unit, and correction processing is performed on the read image data of the original by the CMOS line sensor. Linearity correction means;
An image reading apparatus comprising:

Images