JP6992521B2 - Photoelectric conversion device, defect pixel determination method and image forming device - Google Patents

Description

The present invention relates to a photoelectric conversion device, a method for determining defective pixels, and an image forming device.

A linear image sensor formed of a CCD (Charge Coupled Device) or CMOS (Complementary Metal Oxide Semiconductor) or the like contains defective pixels that cause an abnormality in the saturated charge amount or an abnormality in the dark current of a photodiode (PD). It has been known. A defective pixel has a higher pixel value or a lower pixel value than a normal pixel, resulting in deterioration of image quality. Therefore, it is necessary to detect defective pixels and correct the pixel values.

In Patent Document 1 (Japanese Unexamined Patent Publication No. 2014-110622), it is determined whether or not the pixel values of the target pixels obtained at two different exposure times are linear with respect to time, and the exposure time and the pixel values are not linear. Is disclosed as an image processing apparatus that detects a defect pixel.

However, the image processing apparatus disclosed in Patent Document 1 uses an image at the time of exposure for detecting defective pixels. Since the change in the pixel value due to light is larger than the change in the pixel value due to the dark current, there is a problem that it becomes difficult to detect defective pixels having an abnormality in the dark current.

Further, the pixel value includes an offset component of the circuit. The offset component of the circuit is of a magnitude that cannot be ignored with respect to dark current. Therefore, the image processing apparatus of Patent Document 1 that detects defective pixels using linearity as a determination condition has a problem that it is difficult to accurately detect defective pixels having an abnormality in dark current.

The present invention has been made in view of the above-mentioned problems, and an object of the present invention is to provide a photoelectric conversion device capable of accurately detecting defective pixels of an image sensor, a method for determining defective pixels, and an image forming device.

In order to solve the above-mentioned problems and achieve the object, the present invention uses a photoelectric conversion element that generates an image signal according to the amount of received light and a pixel value obtained in a state where no light is incident on the photoelectric conversion element. A certain dark pixel value is acquired at least in the first time and a second time longer than the first time, and each dark pixel value acquired in the first time and the second time are acquired. By taking the difference between each pixel value in the dark, the noise detection unit that detects the amount of noise in the dark and the pixel whose difference has a noise amount equal to or greater than a predetermined first threshold are determined as defective pixels. Has a part.

According to the present invention, there is an effect that defective pixels of an image sensor can be detected with high accuracy.

FIG. 1 is a cross-sectional view of the MFP of the embodiment. FIG. 2 is a cross-sectional view of a reading device provided in the MFP of the embodiment. FIG. 3 is a hardware configuration diagram of the MFP of the embodiment. FIG. 4 is a block diagram of a photoelectric conversion element provided in the MFP of the embodiment. FIG. 5 is a circuit diagram of a photoelectric conversion element provided in the MFP of the embodiment. FIG. 6 is a diagram for explaining image noise generated in a scanned image due to defective pixels. FIG. 7 is a diagram showing image noise generated in a scanned image when defective pixels are generated in the linear sensor. FIG. 8 is a diagram for explaining a defect pixel detection method as a reference example. FIG. 9 is a diagram for explaining a problem of a defect pixel detection method as a reference example. FIG. 10 is a block diagram of an image reading unit provided in the MFP of the embodiment. FIG. 11 is a block diagram of a noise detection unit provided in the image reading unit. FIG. 12 is a diagram showing pixel values from which circuit offset components have been removed by a noise detection unit provided in the MFP of the embodiment. FIG. 13 is a block diagram of an image reading unit when a black shading correction unit is provided in the MFP of the embodiment. FIG. 14 is a diagram showing an example of defective pixels detected by a noise detection unit and a determination unit provided in the MFP of the embodiment. FIG. 15 is a diagram for explaining an operation of storing defective pixels detected by a noise detection unit and a determination unit based on a predetermined priority in the MFP of the embodiment. FIG. 16 is a flowchart for explaining the detection timing of defective pixels at the time of starting the MFP of the embodiment. FIG. 17 is a flowchart for explaining the detection timing of defective pixels at the time of reading a document using a pressure plate in the MFP of the embodiment. FIG. 18 is a flowchart for explaining the detection timing of defective pixels at the time of reading a document using ADF in the MFP of the embodiment. FIG. 19 is a block diagram of a defect pixel correction unit provided in the MFP of the embodiment. FIG. 20 is a diagram for explaining the defect pixel correction process of the MFP of the embodiment. FIG. 21 is a flowchart showing a flow of defect pixel correction processing of the MFP of the embodiment.

First, the fields of application will be explained. The photoelectric conversion device and the method for determining defective pixels can be applied not only to a device that reads an image but also to a device that senses the presence or absence of light and performs predetermined information processing. Specifically, the photoelectric conversion device and the photoelectric conversion method are on a linear sensor of a multifunction device (MFP: Multifunction Peripheral), a line sensor for autofocus of a camera device or a video camera device, and an interactive whiteboard device (electronic blackboard). It can be applied to a line sensor or the like that reads characters, symbols, or figures written in. Hereinafter, as an example, a photoelectric conversion device and an MFP to which a photoelectric conversion method is applied will be described.

(MFP configuration)
First, FIG. 1 shows a view of the MFP of the embodiment as viewed from the side. FIG. 1 is a perspective view of the main body of the MFP. As shown in FIG. 1, the MFP has a reading device 1 and a main body 2. The reading device 1 has an automatic document feeder (ADF: Auto Document Feeder) 3 and a scanner mechanism 4.

Inside the main body 2, a tandem image forming unit 5, a resist roller 7 that supplies recording paper to the image forming unit 5 from a feeding unit 13 via a transport path 6, an optical writing device 8, a fixing transport unit 9, and a fixing transport unit 9 are provided. , Has a double-sided tray 10. In the image forming unit 5, four photoconductor drums 11 corresponding to four colors of yellow (Y), magenta (M), cyan (C), and black (K) are arranged side by side. An image forming element including a charger, a developer 12, a transfer device, a cleaner, and a static eliminator is arranged around each photoconductor drum 11. Further, between the transfer device and the photoconductor drum 11, an intermediate transfer belt 14 stretched between the drive roller and the driven roller while being sandwiched between the nips of both is provided.

In the tandem image forming apparatus configured as described above, light is written on the photoconductor drum 11 corresponding to each color of YMCK, and the toner of each color is developed by the developer 12, for example, Y, M, C, K. The primary transfer is performed on the intermediate transfer belt 14 in the order of. Then, a full-color image in which four colors are superimposed by the primary transfer is secondarily transferred to a recording paper, and then fixed and ejected. As a result, a full-color image is formed on the recording paper.

(Configuration of ADF and scanner mechanism)
FIG. 2 is a cross-sectional view of the ADF 3 and the scanner mechanism 4. The scanner mechanism 4 includes a contact glass 15 on which a document is placed on the upper surface. Further, the scanner mechanism 4 includes a first carriage 18 including a light source 16 for exposing a document and a first reflection mirror 17, and a second carriage 24 including a second reflection mirror 19 and a third reflection mirror 20. There is. Further, the scanner mechanism 4 includes a lens unit 22 for forming an image of the light reflected by the third reflection mirror 20 on the light receiving region of the photoelectric conversion element 21. Further, the scanner mechanism 4 includes a reference white plate 23 for various distortion corrections by a reading optical system and the like, and a sheet-through reading slit 24. The scanner mechanism 4 receives the reflected light from the document illuminated by the irradiation light from the light source 16 by the photoelectric conversion element 21, converts it into an electric signal (image data), and outputs the light.

The ADF 3 is connected to the main body 2 via a hinge member or the like (not shown) so as to be openable and closable with respect to the contact glass 15. The ADF 3 includes a document tray 28 on which a document bundle 27 composed of a plurality of documents can be placed. Further, the ADF 3 is a separate feeding unit including a feeding roller 29 that separates the originals one by one from the original bundle 27 placed on the original tray 28 and automatically feeds them toward the sheet-through reading slit 25. Also equipped.

(Original scanning operation)
Such a reading device 1 has a scan mode for reading a document placed on a contact glass 15 and a sheet-through mode for reading a document automatically fed by the ADF 3. Before reading the image in the scan mode or the sheet-through mode, the reference white plate 23 is illuminated by the lit light source 16 and the image by the reflected light is read by the photoelectric conversion element 21. Then, shading correction data is generated and stored so that the level of each pixel of the image data for one line becomes a uniform level. The stored shading correction data is used for shading correction of the image data read in the scan mode or the sheet-through mode described below.

In the scan mode, the first carriage 18 and the second carriage 24 are moved in the arrow A direction (sub-scanning direction) by a stepping motor (not shown) to scan the document. At this time, in order to keep the optical path length from the contact glass 15 to the light receiving region of the photoelectric conversion element 21 constant, the second carriage 24 moves at a speed of 1/2 of that of the first carriage 18.

At the same time, the image surface, which is the lower surface of the document placed on the contact glass 15, is illuminated (exposed) by the light source 16 of the first carriage 18. Then, the reflected light from the image plane is sequentially reflected by the first reflection mirror 17 of the first carriage 18, the second reflection mirror 19 of the second carriage 24, and the third reflection mirror 20. Then, the light flux reflected by the third reflection mirror 20 is focused by the lens unit 22 and imaged on the light receiving region of the photoelectric conversion element 21. The photoelectric conversion element 21 photoelectrically converts the light received for each line to generate image data. The image data is digitized, gain adjusted, and output. The original whose image has been read is discharged to an ejection port (not shown).

In the seat-through mode, the first carriage 18 and the second carriage 24 move to the lower side of the seat-through reading slit 25 and stop. After that, the lowermost originals of the original bundle 27 placed on the original tray 28 of the ADF3 are automatically fed by the feeding roller 29 in the arrow B direction (secondary scanning direction), and the sheet-through reading slit 25 is automatically fed. When the document passes through the position, the document is scanned.

At this time, the lower surface (image surface) of the automatically fed document is illuminated by the light source 16 of the first carriage 18. Then, the reflected light from the image plane is sequentially reflected by the first reflection mirror 17 of the first carriage 18, the second reflection mirror 19 of the second carriage 24, and the third reflection mirror 20. Then, the light flux reflected by the third reflection mirror 20 is focused by the lens unit 22 and imaged on the light receiving region of the photoelectric conversion element 21. The photoelectric conversion element 21 photoelectrically converts the light received for each line to generate image data. The image data is digitized, gain adjusted, and output. The original whose image has been read is discharged to a paper ejection port (not shown).

(Hardware configuration of MFP)
Next, FIG. 3 shows the hardware configuration of the MFP. As shown in FIG. 3, the MFP includes a CPU 41, a ROM 42, a RAM 43, an HDD (hard disk drive) 44, and a flash memory 45. Further, the MFP is a network such as a FAX modem 46, an operation panel 47, an engine 48, an ADF 49 (corresponding to ADF 3 shown in FIGS. 1 and 2), a connection interface (connection I / F) 50, an image reading unit 52, and the Internet. It has a communication I / F 51 that performs wired communication or wireless communication via 40. The CPU 41 to the image reading unit 52 are connected to each other via the system bus 18 shown in FIG.

The CPU 41 comprehensively controls the operation of the MFP. The CPU 41 controls the operation of the entire MFP by executing a program stored in the ROM 42 or the HDD 44 using the RAM 43 as a work area (work area), and has various functions such as a copy function, a scanner function, a facsimile function, and a printer function. To realize.

The engine 48 is hardware that performs general-purpose information processing and processing other than communication in order to realize a copy function, a scanner function, a printer function, and the like. The engine 48 includes, for example, a scanner that scans and reads characters and images such as a document or a business card, a plotter that prints on a sheet material such as paper, and the like. The FAX modem 46 has a facsimile communication function for performing facsimile communication.

(Structure of photoelectric conversion element)
FIG. 5 is a detailed block diagram of the photoelectric conversion element 21. As an example, as shown in FIG. 5, the photoelectric conversion element 21 is a linear sensor in which pixels are arranged one-dimensionally along the main scanning direction. The direction in which the pixels are arranged one-dimensionally is the main scanning direction, and the direction two-dimensionally orthogonal to the main scanning direction is the sub-scanning direction.

That is, the photoelectric conversion element 21 is a pixel signal generation circuit 61 formed by arranging pixels one-dimensionally together with a timing signal generation unit 60, and an amplifier (PGA: programmable gain amplifier) that amplifies each image signal with a predetermined gain. It has 62 and an analog-to-digital converter (ADC) 63 that digitizes an image signal from each PGA.

Further, the photoelectric conversion element 21 has a parallel / serial conversion unit 64 that converts an image signal supplied in parallel from the ADC 63 into a serial image signal and supplies it to the subsequent processing unit. The output timing of the pixel signal generation circuit 61 to the parallel / serial conversion unit 64 is controlled based on the clock signal from the timing signal generation unit 60.

When the photoelectric conversion element 21 is a color-compatible photoelectric conversion element, the pixel signal generation circuit 61 is provided for channels of each color such as red, green, blue (RGB) or yellow, magenta, and cyan (YMC). .. Further, PGA62 and ADC63 are also provided for each color channel.

(Circuit configuration of pixel signal generation circuit)
FIG. 5 shows a circuit diagram of a portion corresponding to each pixel of the pixel signal generation circuit 61. As shown in FIG. 5, each pixel of the pixel signal generation circuit 61 has a photodiode PD which is a light receiving element, and a floating diffusion FD which converts the accumulated charge corresponding to the amount of received light into a voltage. Further, each pixel has a charge transfer switch TX that transfers the charge accumulated in the photodiode PD to the floating diffusion FD, and a reset switch RT that resets the potential of the floating diffusion FD to the reset potential A VDD_RT. Further, each pixel is a switch SL for transferring the voltage-converted image signal (Pix_out) to the PGA62 in the subsequent stage and the source follower SF inserted and connected between the power supply voltage A VDD_PIX of the source follower SF and the current source DRG. have.

Since the threshold voltage of the transistor of the source follower SF varies, a circuit offset that does not depend on the amount of incident light occurs in the output signal. The circuit offset causes FPN (Fixed pattern noise) in the image and causes deterioration of image quality.

Further, in the manufacture of the photoelectric conversion element 21, a pixel having an abnormally large or small dark current of the photodiode PD is generated with a certain probability due to heavy metal contamination or crystal defects. Pixels in which the dark current of the photodiode PD is abnormally large are called "white scratch pixels". Further, a pixel having a small capacitance of the photodiode PD and a small amount of saturation charge is generated. Such pixels are called "black scratch pixels". White scratch pixels and black scratch pixels appear on the image as image noise at points having different pixel values from normal pixels, resulting in deterioration of image quality. Further, the dark current noise component of the photodiode PD increases in proportion to time. Further, the dark current noise component of the photodiode PD is temperature-dependent, and although it is an example, it doubles when the temperature rises at 10 ° C.

(Image noise caused by defective pixels)
FIG. 6 shows a state in which image noise is generated when an image is read under conditions in which the incident light intensity and the accumulation time are different with a 9-pixel area sensor including white scratch pixels 301 and black scratch pixels 302. .. The horizontal axis of FIG. 6 is the light accumulation time of the photodiode PD, and the vertical axis is the incident intensity of light with respect to the photodiode PD.

As shown in FIG. 6A or FIG. 6B, the white scratch pixel 301 does not become image noise because the signal component due to the light becomes large and the influence of the dark current becomes inconspicuous when the amount of incident light is large. .. However, as shown in FIG. 6 (c) or FIG. 6 (d), when the amount of incident light is small, the pixel value is higher than that of a normal pixel, so that it appears as image noise at a point brighter than the surroundings. Further, as shown in FIG. 6D, when the accumulation time is long, more conspicuous image noise appears.

As shown in FIG. 6A or FIG. 6B, the black scratch pixel 302 appears as image noise at dark spots because the pixel value is lower than that of a normal pixel when the amount of incident light is large. .. Further, as shown in FIG. 6B, when the accumulation time is long, more noticeable image noise appears. As shown in FIG. 6C or FIG. 6D, the black scratch pixel 302 does not cause image noise because the photodiode PD is not saturated when the amount of incident light is small.

As described above, the white scratch pixel 301 causes the image quality to deteriorate in the dark image, and the black scratch pixel 302 causes the image quality to deteriorate in the light image. In order to prevent these deteriorations in image quality, it is necessary to detect and correct defective pixels. The black scratch pixel 301 is easy to detect because, for example, if image data when exposed for a long time is acquired, normal pixels appear as dark spots in a saturated white image.

(Image noise caused by defective pixels of the linear sensor)
FIG. 7A is a diagram showing the positions of the white scratch pixels on the linear sensor, and the pixels indicated by the diagonal lines are the white scratch pixels. FIG. 7B is a diagram showing a dark image (an image acquired in a state where the linear sensor is not exposed) acquired by a linear sensor including white scratch pixels. The pixel value of the white scratch pixel is higher than that of the normal pixel. Therefore, if white scratches are included on the linear sensor, as shown in FIG. 7B, the image (pixels) at the position corresponding to the white scratch pixels in the entire image is larger than the surrounding image. Vertical streaks that become brighter Image noise appears.

Further, when the linear sensor includes black scratch pixels, the opposite result is obtained as compared with the case where the above-mentioned white scratch pixels are included. That is, the pixel value of the black scratch pixel is lower than that of the normal pixel. Therefore, when the linear sensor includes black scratch pixels, vertical streak image noise appears in which the image (pixel) at the position corresponding to the black scratch pixel in the entire image becomes darker than the surrounding image.

As described above, in the case of the linear sensor, since the adverse effect of the defective pixel appears as vertical streak image noise, the adverse effect of the defective pixel becomes larger than that of the area sensor in which the adverse effect of the defective pixel appears as the image noise of the point. The adverse effect of the defective pixel becomes more remarkable when a plurality of defective pixels are present adjacent to each other.

(Reference example of defect pixel detection method)
Next, a reference example of the defect pixel detection method will be described. 8 (a) and 8 (b) are diagrams showing pixel values when the same subject is imaged at different exposure times t1 and t2 (t2> t1). Of these, FIG. 8A is a pixel value when the subject is imaged with normal pixels. Further, FIG. 8B is a pixel value when the subject is imaged with black scratch pixels.

Assuming that the pixel value S = 0 and the pixel value S changes linearly with time when the exposure time t = 0, the rate of change S1 / t1 with respect to time is calculated from the pixel value S1 at the exposure time t1 and then exposed. The ideal pixel value S2 at time t2 is calculated. The calculation formula of the pixel value S2 is shown in the following equation (1).

S2 = (S1 / t1) × t2 ... (Equation 1 formula)

In reality, electrical noise is included with respect to the pixel value. Therefore, it is considered that the pixel value S3 at the exposure time t2 has an error (fixed value) of ± α with respect to the pixel value S2, and it is determined whether or not the pixel value S3 is within the range of the pixel value S2 ± α. By doing so, it is determined whether or not the pixel is defective. The dark current noise component increases in proportion to time in the same manner as the signal component. Further, the circuit offset component is a fixed value individually possessed by each pixel of the photoelectric conversion element. However, since the dark current noise component and the circuit offset component have small values with respect to the signal component, they do not affect the determination of defective pixels.

In the case of normal pixels, the signal component linearly changes with respect to the exposure time as shown in FIG. 8 (a). Therefore, since the pixel value S3 is within the range of the pixel value S2 ± α, it is determined to be a normal pixel. On the other hand, in the case of a black scratch pixel, since the amount of saturated charge is small, the charge accumulated at a low pixel value is saturated at the exposure time t2 as shown in FIG. 8 (b). Therefore, the value of the pixel value S3 is smaller than the pixel value S2 of the normal pixel. Therefore, since the pixel value S3 does not fall within the range of the pixel value S2 ± α, it is determined to be a defective pixel.

(Problems with the defect pixel detection method in the reference example)
8 (a) and 8 (b) are examples of black scratch pixels, whereas FIGS. 9 (a) and 9 (b) are the same with different exposure times t1 and t2 (t2> t1). It is a figure which shows the pixel value of the white scratch pixel when the subject is imaged. Of these, FIG. 9A shows a pixel value obtained when the white scratched pixel is irradiated with light, and FIG. 9B shows a case where the white scratched pixel is not irradiated with light (in the dark). ) Indicates the pixel value obtained.

As shown in FIG. 9A, in the case of white scratch pixels, the dark current noise component increases with the exposure time. However, since the dark current noise component is small with respect to the signal component due to the incident light, the pixel value S3 of the white scratch pixel is within the range of the pixel value S2 ± α. Therefore, there is a problem that the white scratched pixel is erroneously determined as a normal pixel.

Further, as shown in FIG. 9B, the pixel value of the white scratch pixel in darkness has a circuit offset component larger than that of the dark current noise component. Therefore, when the pixel value S = 0 when the exposure time t = 0 and the pixel value S is assumed to change linearly with time, the defective pixel is determined based on the pixel value S2, even if it is a normal pixel. , The pixel value S3 does not fall within the range of the pixel value S2 ± α, which causes an inconvenience that a normal pixel is erroneously determined as a defective pixel. Therefore, the defect pixel detection method in the reference example has a problem that it is difficult to detect white scratch pixels even if black scratch pixels can be detected.

(Structure and operation of the main part of the MFP of the embodiment)
In the MFP of the embodiment, the image reading unit 52 has a configuration shown in FIG. 10 in order to accurately detect such defective pixels. That is, in the MFP of the embodiment, the image reading unit 52 includes a light-shielding unit 71 that shields light from incident on the photoelectric conversion element 21, a noise detection unit 72 and a determination unit 83 that detect defective pixels, and noise. It has a defect pixel correction unit 73 that corrects a defect pixel detected by the detection unit 72 and the determination unit 83.

As an example, the light-shielding unit 71 is a light-shielding mechanism in which the light-receiving surface of the photoelectric conversion element 21 is covered with a light-shielding member that does not transmit light so that light is not incident on the photoelectric conversion element 21. The ADF3, the pressure plate, or the reference white plate 23 may be used for shading the photoelectric conversion element 21.

The CPU 41 controls the light-shielding portion 71 so that the photoelectric conversion element 21 is covered with the light-shielding member when the defective pixel is corrected. The photoelectric conversion element 21 has the above-mentioned accumulation times t1 and t2 (t2) in "dark time", which is the time when the photoelectric conversion element 21 is covered with the light-shielding member (the time when light is not incident on the photoelectric conversion element 21). > Acquire the pixel data of t1).

The noise detection unit 72 and the determination unit 83 detect defective pixels based on the pixel data acquired from the photoelectric conversion element 21. The defect pixel correction unit 73 determines the pixel data of the defective pixel among the pixel data supplied from the photoelectric conversion element 21 based on the detection result of the defect pixel detected by the noise detection unit 72 and the determination unit 83. Is corrected and supplied to an external device such as a monitor device or a storage device.

Although the noise detection unit 72, the determination unit 83, and the defect pixel correction unit 73 will be described as being realized by hardware, one or all of them may be realized by software. In this case, for example, the CPU 41 reads out the defect pixel correction program stored in the HDD 44 or the flash memory 45 shown in FIG. 3, and deploys the noise detection unit 72, the determination unit 83, and the defect pixel correction unit 73 in the RAM 43. .. Then, as described below, the CPU 41 executes the defect pixel detection process and the defect pixel correction process.

(Configuration and operation of noise detection unit and judgment unit)
FIG. 11 shows a block diagram of the noise detection unit 72. As shown in FIG. 11, when the noise is detected, the noise detection unit 72 has a memory 81 that stores the pixel value St1x of the charge accumulation time t1 in the dark, a pixel value St1x stored in the memory 81, and darkness. It has a subtractor 82 for detecting the difference between the accumulation time t2 (t2> t1) and the pixel value St2x at the time. The configuration for detecting the difference between the pixel value St1x and the pixel value St2x is not limited to the subtractor. For example, the same result can be obtained by the configuration of a multiplier that inverts the pixel value St1x to a negative value and an adder that adds the multiplication result and the pixel value St2x. Further, as the memory 81, for example, the RAM 43 or the flash memory 45 shown in FIG. 3 provided outside the noise detection unit 72 may be used.

As an example, in the case of the MFP of the embodiment, the photoelectric conversion element 21 for each color is provided like RGB or YMC. Therefore, each RGB pixel data (pixel value) from the photoelectric conversion element 21 for each color is supplied to the noise detection unit 72. The noise detection unit 72 stores in the memory 81 the pixel values St1x for all the pixels of the photoelectric conversion element 21 for each color acquired at the accumulation time t1 in the dark when noise detection is performed.

In this example, the pixel value St1x having the storage time t1 is stored in the memory 81, but the pixel value St2x for each color acquired in the storage time t2 longer than the storage time t1 may be stored in the memory 81. ..

When the pixel value St2x for each color acquired in the storage time t2 longer than the storage time t1 is supplied to the subtractor 82 in the dark when the CPU 41 detects defective pixels, the CPU 41 has the same pixel position as each pixel value St2x. The pixel value St1x at the pixel position is read from the memory 81 and supplied to the subtractor 82. The subtractor 82 calculates the difference value St2x-St1x for each color.

The determination unit 83 compares the difference value St2x-St1x at each pixel position with the predetermined threshold value Dth, and determines as a defective pixel a pixel having a pixel value in which the difference value St2x-St1x is equal to or greater than the threshold value Dth. Further, if at least one of the pixel values of each color at the same pixel position has a pixel value equal to or higher than the threshold value Dth, the determination unit 83 determines all the color pixels at the same pixel position as defective pixels (first). Judgment processing). The threshold value Dth may be a fixed value, or may be set by a user or the like to an arbitrary value.

The operation of determining such a defective pixel will be specifically described. Each pixel value of the accumulation time t1R, t1G, t1B shown in FIG. 12A shows the pixel value of the normal pixel of each color in the accumulation time t1, and each pixel value of the accumulation time t2R, t2G, t2B is the accumulation time t2. The pixel value of the normal pixel of each color in is shown. When the pixel values of the storage times t1R, t1G, and t1B are subtracted from the pixel values of the storage times t2R, t2G, and t2B, the circuit offset component that is a substantially fixed value is removed regardless of the passage of time, and the dark current noise. Each pixel value, which is the value of the component, is generated. In the case of normal pixels, each pixel value indicating the value of the dark current noise component of each pixel is less than the threshold value Dth as shown in FIG. 12 (c). In this case, the determination unit 83 determines the pixels of each color as normal pixels.

On the other hand, the pixel values of the accumulation times t1R, t1G, and t1B shown in FIG. 12B indicate the pixel values of the normal pixels of each color at the accumulation time t1, and the pixel values of the accumulation times t2R, t2G, and t2B. Indicates the pixel value of the defective pixel of each color at the accumulation time t2. As described above, when the pixel values of the storage times t1R, t1G, and t1B are subtracted from the pixel values of the storage times t2R, t2G, and t2B, the circuit offset component that is a substantially fixed value is removed regardless of the passage of time. And each pixel value which is the value of the dark current noise component is generated.

However, in the case of a white scratch pixel, a high dark current noise component appears even in the dark when no light is incident, as in the pixel value of the accumulation time t2G in FIG. 12 (b). The example of FIG. 12B is an example in which the pixel for G is a white scratch pixel. Therefore, among the dark current noise components of the pixels of each color generated by subtracting the pixel values of the storage times t1R, t1G, and t1B from the pixel values of the storage times t2R, t2G, and t2B, FIG. 12 (c) As shown in the above, a pixel having a dark current noise component having a threshold value Dth or more is detected.

In this case, the determination unit 83 determines the pixel of the color having the dark current noise component having the threshold value Dth or more as the defective pixel, and also determines the pixel of another color at the same pixel position as the defective pixel. In other words, when the determination unit 83 determines even one as a defective pixel, the determination unit 83 determines the pixels of all colors at the same pixel position as defective pixels.

The temperature of the photoelectric conversion element 21 rises due to heat generated by driving. Therefore, the longer the time elapses from the start of driving, the higher the temperature. Since the dark current noise component increases as the temperature rises, the white scratch pixel is determined based on the difference St2x-St1x between the pixel value St2x having a long storage time t2 and the pixel value St1x having a short storage time t1. By doing so, white scratch pixels can be detected with high accuracy.

(Second judgment process)
Further, as described above, when all the pixel values of each color at the same pixel position are less than the threshold value Dth, the determination unit 83 determines as a normal pixel, but further performs the second determination process described below. You may.

That is, when the determination unit 83 determines that all the pixel values of each color at the same pixel position are less than the threshold value Dth, the determination unit 83 integrates all the pixel values of each color at the same pixel position before determining the normal pixel. It is determined whether or not the integrated value is equal to or higher than the predetermined threshold value Dth2. Then, when the integrated value of the pixel values of each color is equal to or higher than the predetermined threshold value Dth2, the pixels of all the colors at the same pixel position are determined as defective pixels (second determination process).

The determination unit 83 may perform only the first determination process to determine the defective pixel, but the determination accuracy of the defective pixel may be improved by performing the second determination process together with the first determination process. Can be done.

(Black shading correction processing)
Here, the image reading unit 52 may be provided with a black shading correction unit 88 as shown in FIG. In this case, the black shading correction unit 88 is provided between the photoelectric conversion element 21 and the noise detection unit 72. The example of FIG. 13 is an example in which the black shading correction unit 88 is provided as hardware. However, the black shading correction unit 88 described below may be realized by software by the CPU 41 executing, for example, a black shading correction program.

The black shading correction unit 88 acquires the pixel values of all the pixels acquired from the photoelectric conversion element 21 in the dark. The black shading correction unit 88 stores the acquired pixel values in the dark in a storage unit provided inside the black shading correction unit 88 or outside the black shading correction unit 88. Then, the black shading correction unit 88 is stored in the storage unit in the dark from each pixel value at light time, which is image data obtained at the time of scanning by incident light on the photoelectric conversion element 21 to read a document or the like. Each pixel value of is subtracted.

As a result, the circuit offset component which is fixed pattern noise (FPN: Fixed Pattern Noise) can be removed from the image pickup data, and the S / N ratio (Signal to Noise Ratio) of the image pickup data can be improved.

Further, each pixel value in the dark stored in the storage unit for black shading correction can be used in the above-mentioned white scratch pixel determination process by the noise detection unit 72 in the subsequent stage. Therefore, when the black shading correction unit 88 is provided in the image reading unit 52, the noise detection memory 81 can be omitted. In other words, the storage unit for black shading correction and the storage unit for noise detection can be used in combination, and the number of parts can be reduced and the configuration can be simplified.

(Memory operation of judgment result)
Next, the operation of storing the determination result from the determination unit 83 will be described. 14 (a) and 14 (b) are histograms that are the determination results of defective pixels when the number of pixels of the photoelectric conversion element 21 for each color of RGB, which is a linear sensor, is 20 pixels, respectively. be. Of these, FIG. 14A is a histogram that is the N-1th determination result, and FIG. 14B is a histogram that is the Nth determination result. "N" is a natural number of 2 or more. Further, in FIGS. 14 (a) and 14 (b), the histogram of the diagonal line pattern is the histogram of the pixel of R, the histogram of the dotted line pattern is the histogram of the pixel of G, and the histogram of the white line is the histogram of B. It is a histogram of pixels.

In the case of the examples of FIGS. 14 (a) and 14 (b), for example, a pixel having a pixel value of the first threshold value Dth = 8 or more is determined to be a defective pixel. That is, in the pixel of R of the first pixel, the value of the above-mentioned "pixel value St2x of accumulation time t2-pixel value St1x of accumulation time t1" is a pixel value equal to or greater than the first threshold value Dth. In this case, the determination unit 83 determines each of the RGB pixels of the first pixel as a defective pixel. If even one of RGB is determined to be a defective pixel, the defect pixel correction process can be performed with high accuracy by determining the three RGB pixels as defective pixels. Similarly, in the pixel G of the fourth pixel, the value of the above-mentioned "pixel value St2x of accumulation time t2-pixel value St1x of accumulation time t1" is a pixel value equal to or greater than the first threshold value Dth. Therefore, the determination unit 83 determines each of the fourth RGB pixels as defective pixels.

The determination unit 83 stores, for example, the determination information for 6 pixels among the determination information indicating the determination result of each pixel in a storage unit such as the RAM 43, the HDD 44 shown in FIG. 3 or the memory 81 shown in FIG. .. 15 (a) to 15 (d) are schematic views of determination information stored in the storage unit. Of these, FIG. 15 (a) is a schematic diagram of the determination information that is the N-1th determination result, and FIGS. 15 (b) to 15 (d) are schematic diagrams of the determination information that is the Nth determination result, respectively. It is a figure. N is a natural number of "2" or more.

As shown in FIG. 15A, the determination unit 83 stores the rank (priority), the pixel position, the maximum value of the noise amount of each pixel of RGB, and the maximum value of the noise amount of each pixel of RGB as determination information. Remember in the department. The example of FIG. 15A is an example in which the determination information of each pixel is stored in descending order of the amount of noise. Specifically, in the case of the example of FIG. 15A, it is determined that each pixel at the pixel positions of 13, 7, 19, 4, 17, and 11 has a large amount of noise in order. The pixels at the 13 pixel positions of rank "1" have RGB noise amounts of 7, 13, and 7. Of these, the noise amount of "13" is the maximum noise amount. Similarly, the pixels at the 19 pixel positions of rank "3" have RGB noise amounts of 11, 6, and 6. Of these, the noise amount of "11" is the maximum noise amount.

FIG. 15B shows the determination information that is the Nth determination result in the example of storing the determination information of each pixel in descending order of the amount of noise. That is, the determination information in FIG. 15B is the determination information obtained by the determination unit 83 performing the determination again after the determination information shown in FIG. 15A is obtained. As an example, the determination unit 83 stores up to six determination information in the storage unit. Then, the determination information stored in the storage unit is rewritten according to the determination information obtained later.

Specifically, in the case of the noise amount priority example shown in FIG. 15 (b), as can be seen in comparison with FIG. 15 (a), as the determination information corresponding to the rank "6", RGB at 10 pixel positions. When new determination information having noise amounts of 8, 8 and 9 is obtained, the determination information of rank "6" is rewritten to the determination information of the pixel positions of 10.

The RGB noise amounts of rank "6" shown in FIG. 15 (a) are 8, 8 and 9, whereas the noise amount of rank "6" is newly used as the determination information of rank "6" as shown in FIG. 15 (b). The amount of RGB noise in the determination information of the 10 pixel positions to be rewritten is 8, 8 and 9, which is the same as the determination information of the 11 pixel positions.

Here, the user of the MFP often arranges the image pickup target such as a document in the center of the document mounting table. Therefore, when the noise amount is the same, the determination unit 83 selects the determination information of the pixel position closer to the central pixel of the photoelectric conversion element 21 and stores (rewrites) it in the storage unit. Thereby, more effective determination information can be stored in the storage unit and used for the correction process of the defective pixel.

Further, the above-mentioned example is an example of storing determination information in the case where the noise amount is prioritized. However, when the adjacent pixels are both defective pixels such as the pixels at the pixel position 1 and the pixel position 2 shown in FIG. 14B, the effect on the image quality is large. Therefore, as shown in FIG. 15C, the determination unit 83 may rank the determination information of the adjacent defective pixels and store them in the storage unit. In the example of FIG. 15C, determination information of 10 adjacent pixel positions and 11 pixel positions, respectively, and determination information of 2 adjacent pixel positions and 1 pixel position are preferentially stored. Is.

Further, even if the noise level is equal to or less than that, when a plurality of color channels are defective pixels, the influence on the image quality is large. Therefore, as shown in FIG. 15D, the determination unit 83 may have a plurality of color channels rank the determination information of the defective pixel and store it in the storage unit. In the example of FIG. 15D, the determination information of the pixels at the pixel positions of 15, 13, 7, 19, 4, and 17 is preferentially stored as the pixels of the color channel having a large number of defective pixels. Is.

It should be noted that the ranking may be performed by comprehensively considering the above-mentioned defective pixels having a large amount of noise, adjacent defective pixels, and defective pixels of a plurality of color channels.

(Defect pixel detection processing operation at startup)
Next, the MFP of the embodiment is adapted to detect such defective pixels when the MFP is activated. The flowchart of FIG. 16 shows the flow of an operation of detecting defective pixels at startup. First, when the main power supply of the MFP is turned on, the CPU 41 shown in FIG. 3 applies (ON) a drive voltage to the photoelectric conversion element 21 for warm-up (step S1).

The first carriage 18 and the second carriage 24 of the MFP shown in FIG. 2 are set with, for example, a standby position which is a position when the document is conveyed and read. Therefore, the CPU 41 performs homing control to move the first carriage 18 and the second carriage 24 to the standby position after applying a drive voltage to the photoelectric conversion element 21 (step S2). The state in which the first carriage 18 and the second carriage 24 are waiting at the standby position is an example of the read standby.

Next, the CPU 41 moves and controls the first carriage 18 under the reference white plate 23 for adjustment (step S3). In order to prevent light from leaking to the outside of the MFP, the CPU 41 drives the light source 16 to light after moving the first carriage 18 under the reference color plate 23 (step S4). After driving the light source 16 to turn on and performing adjustment operations for each part such as adjusting the amount of light of the light source 16 and adjusting the gain of the PGA 62 shown in FIG. 4 using the reference color plate 23 (step S5), the light source 16 is turned off. (Step S6).

Next, after the light source 16 is turned off, the CPU 41 confirms whether or not there is an abnormality in the black level (step S7), and after confirming the black level, performs the above-mentioned defect pixel detection process (step S8).

Finally, the CPU 41 performs the homing control described above (step S9), and the activation process of the flowchart of FIG. 16 is completed.

Defect pixels are temperature-dependent, and the higher the temperature of the photoelectric conversion element 21, the higher the detection accuracy. Therefore, by performing the defective pixel detection process at the end of the MFP activation process shown in the flowchart of FIG. 16, defective pixels can be detected with high detection accuracy. Further, in step S8, of the two dark pixel values acquired for detecting defective pixels with different accumulation times, the pixel value having the longer accumulation time is acquired later, so that the difference in the amount of noise between the pixels is increased. Since it becomes large, defective pixels can be detected with high accuracy.

Further, if the first carriage 18 is located below the reference white plate 23, or the pressure plate (or ADF) is closed, and the light source 16 is not turned off, no dark data can be obtained. Therefore, defective pixels are detected during the activation process of the MFP that controls the movement of the first carriage 18 under the reference white plate 23. As a result, the defective pixel detection process can be performed regardless of the open / closed state of the pressure plate (or ADF). Further, it is possible to prevent the inconvenience that the homing control for moving the first carriage 18 is performed for each defect pixel detection process, and it is possible to shorten the waiting time for the user until the original can be read.

Further, when the light source 16 is turned off and turned on, it takes time for the light source 16 to stabilize. Therefore, it is desirable that the number of times the light source 16 is driven to turn off and to be turned on is small. Therefore, the defect pixel detection process (step S8) is performed together with the black level confirmation process (step S7) when the light source 16 is turned off during the startup process of the MFP. As a result, it is possible to prevent the inconvenience that the number of times the light source 16 is driven to turn off and to be turned on increases due to the defect pixel detection process.

As shown in FIG. 13, when the image reading unit 52 has the shading correction unit 88, the noise detection unit 72 is used for the black shading correction process when executing the process of the above step S8. Acquires the pixel value at dark time.

(Detection processing operation of defective pixels when scanning a document using a pressure plate)
FIG. 17 is a flowchart for explaining the operation of detecting defective pixels when reading a document using a pressure plate. In this case, the defective pixel is detected after the image data of the document has been acquired and the light source 16 is turned off.

After reading the document using the pressure plate, there is a timing at which light does not enter the photoelectric conversion element 21 due to shading by the pressure plate or the reference white plate 23, and it is not necessary to individually create a state in which light does not enter the photoelectric conversion element 21. Further, the temperature of the photoelectric conversion element 21 is higher after the image data is acquired because the photoelectric conversion element 21 is driven for a longer time than during the start-up process described with reference to the flowchart of FIG. High-precision defective pixel detection processing becomes possible.

That is, in the flowchart of FIG. 17, when the reading of the original is started using the pressure plate, the CPU 41 applies (ON) a drive voltage to the photoelectric conversion element 21 for warm-up (step S11), and the light source 16 Is driven to light (step S12).

Next, the CPU 41 moves the first carriage 18 in the A direction shown in FIG. 2, acquires image data of the original (step S13, step S14), and controls the light source 16 to be turned off (step S15). Next, the CPU 41 determines whether or not the pressure plate is closed (step S16).

When the pressure plate is closed (step S16: Yes), the CPU 41 performs the above-mentioned defective pixel detection process (step S17) and corrects the defective pixel of the image data detected in step S14 (step S18). .. A specific example of the defect pixel correction process will be described later. Then, the CPU 41 performs the homing control described above in step S19, and ends the processing of the flowchart of FIG.

On the other hand, when the pressure plate is not closed (step S16: No), the CPU 41 performs the homing control described above in step S20. In step S21, the CPU 41 determines whether or not the first carriage 18 is located below the reference white plate 23 by the homing control described above.

When it is determined that the first carriage 18 is located below the reference white plate 23 (step S21: Yes), the CPU 41 performs the above-mentioned defective pixel detection process in step S22. Then, the defective pixel correction process of the image data detected in step S14 is performed (step S23), and the process of the flowchart of FIG. 17 is completed.

When the pressure plate is closed, the photoelectric conversion element 21 is in a light-shielding state, so that defective pixels can be detected immediately after the light source 16 is turned off. At this time, of the two dark image data acquired for detecting defective pixels, which has a longer storage time, the difference in the amount of noise between the pixels becomes larger by making the longer storage time longer than usual, so that the accuracy is correct. As described above, defective pixels can be detected well.

Further, it is preferable that the defect pixel correction process (step S18, step S23) is performed using the latest determination information of the defective pixel. Therefore, the defect pixel correction process (step S18, step S23) is executed after the defect pixel detection process (step S17, step S22). Immediately after the homing control (step S20) is performed, the defective pixel correction process may be performed.

(Operation for detecting defective pixels when scanning a document using ADF)
FIG. 18 is a flowchart for explaining the operation of detecting defective pixels when reading a document using the ADF3. In the flowchart of FIG. 18, when the reading of the document using the ADF 3 is started, the CPU 41 applies (ON) a drive voltage to the photoelectric conversion element 21 for warm-up (step S31), and turns on the light source 16. It is driven to light (step S32).

Next, the CPU 41 starts the transfer control of the document set in the ADF3 (step S33), and acquires the image data of the document (step S34). Then, the CPU 41 uses the determination information of the defective pixel detected at the time of starting the MFP described using the flowchart of FIG. 16 or at the time of reading the document using the pressure plate described using the flowchart of FIG. 17, and steps. The defective pixel correction process of the image data acquired in S34 is performed (step S35).

Next, in step S36, the CPU 41 determines whether or not all the documents set in the ADF 3 have been read. When it is determined that the scanning of all the originals is completed (step S36: Yes), the CPU 41 controls the light source 16 to be turned off (step S37), performs the above-mentioned defective pixel detection process (step S38), and is shown in FIG. End the flow chart processing.

On the other hand, when it is determined that the reading of all the originals has not been completed (step S36: No), the CPU 41 controls the light source 16 to be turned off (step S39), and performs the above-mentioned defective pixel detection process (step S39). Step S40). That is, when the ADF3 is used to continuously read a document, the CPU 41 performs a defect pixel detection process while scanning the document and the next document. As a result, the determination information of the defective pixel can be updated in real time, and the defective pixel can be corrected according to the state. When the defective pixel detection process is completed, the CPU 41 turns on the light source 16 and returns the process to step S34.

In the process of detecting defective pixels in step S40, it takes time for the amount of light to stabilize after the light source 16 is turned on, and the total reading time may be extended. You may go every time.

Further, when scanning a document using ADF3, ADF3 should be closed at the end of scanning. Therefore, in step S37, of the two dark pixel values acquired for detecting defective pixels, the one with the longer storage time is longer than usual, as in the case of scanning the original using the pressure plate. You may get it.

(Correcting operation of defective pixels)
Next, a defect pixel correction processing operation performed using the determination information acquired as described above will be described. FIG. 19 is a detailed block diagram of the defect pixel correction unit 73 of the MFP 1. As shown in FIG. 19, the defect pixel correction unit 73 has a correlation degree calculation unit 101 and a replacement processing unit 102 together with the memory 100.

As shown in FIG. 20 as an example, the defect pixel correction unit 73 uses 16 pixels before and after and 3 lines before and after the defect pixel determined to be the interpolation target by the determination unit 83 as a search area (target pattern) and is a pixel as an interpolation candidate. To explore. Further, the defect pixel correction unit 73 sets a region for three lines formed by a defect pixel, two pixels before and after the defect pixel, and two pixels above and below the two pixels before and after the defect pixel as a template. Then, as shown in FIG. 20, the defect pixel correction unit 73 compares the template with the above-mentioned target pattern (image data of 16 pixels before and after the defect pixel and 3 lines before and after the defect pixel) while moving the template pixel by pixel, and determines the target pattern. Calculate the degree of correlation between each part and the template.

As the degree of correlation, an SAD (Sum of Absolute Difference) value is used as an example. For example, in the case of the example shown in FIG. 20, the defect pixel correction unit 73 calculates the SAD values for 24 pieces based on the following mathematical formula (for example, when the SAD values S1 to S24 are obtained). Then, the defect pixel correction unit 73 replaces the defect pixel with the defect pixel by using the center pixel of the portion of the target pattern for which the smallest SAD value (= maximum correlation degree) is calculated as the replacement pixel.

In the following mathematical formula and FIG. 20, "S" means a SAD value. “S1” indicates the first SAD value, and S24 indicates the 24th SAD value. “P” means a pixel of the target pattern. “P11” means the first pixel of the first line of the target pattern, and “P321” means the 32nd pixel of the first line of the target pattern. Similarly, "P13" means the first pixel of the third line of the target pattern, and "P323" means the 32nd pixel of the third line of the target pattern. Further, "T" means a pixel of the template. “T11” means the first pixel of the first line of the template, and “T41” means the fourth pixel of the first line of the template. Similarly, "P13" means the first pixel of the third line of the target pattern, and "P43" means the fourth pixel of the third line of the target pattern.

S1 = (P11-T11) + (P12-T12) + (P13-T13) + (P21-T21) + (P22-T22) + (P23-T23) + (P31-T31) + (P32-T32) + (P33-T33) ... S24 = (P281-T11) + (P282-T12) + (P283-T13) + (P291-T21) + (P292-T22) + (P293-T23) + (P301-T31) ) + (P302-T32) + (P303-T33)

The defect pixel correction unit 73 calculates the SAD values S1 to S24 by performing arithmetic processing based on such a mathematical formula. Then, for example, when the value of the SAD value S22 is the minimum value, the pixel of P282, which is the center pixel of the portion of the target pattern for which the minimum value is calculated, is used as the replacement pixel and is replaced with the defective pixel.

FIG. 21 is a flowchart showing the flow of such interpolation processing operation. In step S51, the correlation degree calculation unit 101 shown in FIG. 19 sets “1” as the number of operations of the SAD value in the counter (n = 1: n is the number of operations). That is, in the case of the example shown in FIG. 20, the correlation degree calculation unit 101 calculates a total of 24 SAD values. Therefore, in step S51, the correlation degree calculation unit 101 first sets "1", which indicates that it is the first SAD value calculation operation, in the counter. In addition, each time the correlation degree calculation unit 101 completes the calculation of one SAD value, the counter value is set one by one, such as "1"-> "2"-> "3" ... "24". Increment. The timing of increment is the timing of moving the template by one pixel with respect to the target pattern.

Next, in step S52, the correlation degree calculation unit 101 calculates the SAD value (Sn: S means the SAD value, n is the number of the calculated SAD values 1 to 24) based on the above mathematical formula. do. In step S53, the correlation degree calculation unit 101 determines whether or not the number of operations is 24 (n = 24: n is the number of operations). As described above, the correlation degree calculation unit 101 performs the above-mentioned calculation while moving the templates one by one with respect to the target pattern. Therefore, when the number of operations is less than 24 (step S53: No), the process proceeds to step S55, the number of operations is incremented by one (n = n + 1), and the process is returned to step S52. Then, in step S52, the SAD value is calculated again.

When the 24 SAD values are calculated in this way (step S53: Yes), the process proceeds to step S54, and the replacement processing unit 102 shown in FIG. 19 calculates the minimum value among the 24 SAD values. The central pixel of the pattern portion is used as the replacement pixel, and the defective pixel is replaced with the defective pixel (interpolation processing), and the processing of the flowchart of FIG. 21 is completed.

Finally, the embodiments described above are presented as an example and are not intended to limit the scope of the invention. This novel embodiment can be implemented in various other embodiments, and various omissions, replacements, and changes can be made without departing from the gist of the invention. Further, the embodiment and the modification of the embodiment are included in the scope and gist of the invention, and are included in the scope of the invention described in the claims and the equivalent scope thereof.

1 Multifunction device (MFP) reader 2 MFP body 3 Automatic document feeding mechanism (ADF)
4 Scanner mechanism 21 Photoelectric conversion element 41 CPU
60 Timing signal generator 61 Pixel signal generator 62 Amplifier (PGA)
63 Analog-to-digital converter (ADC)
64 Parallel / serial converter 64
71 Shading unit 72 Noise detection unit 73 Defect pixel correction unit 81 Memory 82 Subtractor 83 Judgment unit 88 Black shading correction unit

Japanese Unexamined Patent Publication No. 2014-110622

Claims (17)

A photoelectric conversion element that generates an image signal according to the amount of received light,
The pixel value in the dark, which is the pixel value obtained when no light is incident on the photoelectric conversion element, is used.
Each pixel value in the dark acquired in at least the first time and the second time longer than the first time and acquired in the first time and each in the dark acquired in the second time. A noise detection unit that detects the amount of noise in the dark by taking the difference between the pixel values,
A photoelectric conversion device having a determination unit for determining a pixel having the noise amount whose difference is equal to or greater than a predetermined first threshold value as a defective pixel. A claim characterized by further comprising a correction unit that corrects the pixel value of the pixel determined as the defective pixel among the pixel values of each pixel of the photoelectric conversion element to the pixel value of a normal pixel and outputs the value. Item 1. The photoelectric conversion device according to Item 1. The photoelectric conversion device according to claim 1 or 2, further comprising a light-shielding portion that shields light from incident on the photoelectric conversion element. As the photoelectric conversion element, a photoelectric conversion element for a plurality of predetermined colors is provided.
When the determination unit determines that the pixel of the photoelectric conversion element for at least one color among the pixels of the photoelectric conversion element for each color is a defective pixel, the determination unit corresponds to the same pixel position as the pixel determined as the defective pixel. The photoelectric conversion device according to any one of claims 1 to 3, wherein the pixels of the photoelectric conversion element for other colors are also determined as defective pixels. When the pixels of the photoelectric conversion element for each color corresponding to the same pixel position are less than the first threshold value, the determination unit of the photoelectric conversion element for each color corresponding to the same pixel position. The integrated value of the pixel values is calculated, and when the integrated value is equal to or higher than a predetermined second threshold value, the pixels of the photoelectric conversion element for each color corresponding to the same pixel position are determined as defective pixels. 4. The photoelectric conversion device according to claim 4. Shading correction of each pixel value at the time of imaging is performed by subtracting each pixel value at the time of darkness stored in the storage unit from each pixel value at the time of imaging obtained by incident light on the photoelectric conversion element. Further equipped with a black shading correction unit for processing,
The noise detection unit detects the amount of noise by using each of the dark pixel values stored in the storage unit as each of the dark pixel values acquired in the second time or the first time. The photoelectric conversion device according to any one of claims 1 to 5, wherein the above-mentioned is performed. The determination unit is characterized in that the determination information regarding the pixel determined as the defective pixel is stored in the storage unit according to a predetermined priority, and the stored determination information is rewritten according to the priority. The photoelectric conversion device according to any one of claims 6. The photoelectric conversion device according to claim 7, wherein the determination unit stores the determination information of the defective pixels in the storage unit in descending order of the amount of noise. The photoelectric conversion device according to claim 7, wherein the determination unit stores the determination information of adjacent defective pixels in the storage unit in a predetermined high order. As the photoelectric conversion element, a photoelectric conversion element for a plurality of predetermined colors is provided.
The determination unit is characterized in that, among the color channels in which a predetermined number of pixels for each color are grouped together, the determination information of the color channel having many defective pixels is stored in the storage unit in a predetermined high order. The photoelectric conversion device according to claim 8 or 9. The determination unit of the defective pixel in one or both of the time after the start-up process of the device to which the photoelectric conversion device is applied, or when the reading operation of the photoelectric conversion element is completed and the darkness is set. The photoelectric conversion device according to any one of claims 1 to 10, wherein the determination is performed. Any of claims 1 to 11, wherein the determination unit determines the defective pixel each time the photoelectric conversion element is read once or every time a plurality of readings are completed. The photoelectric conversion device according to claim 1. The claim is characterized in that the noise detection unit acquires the pixel value acquisition order in the dark for detecting the noise amount in the second time after the acquisition in the first time. The photoelectric conversion device according to any one of items 1 to 12. The photoelectric conversion device according to claim 2, wherein the correction unit executes the correction process while waiting for reading of the photoelectric conversion element. The correction unit arranges a template of a partial size of the target pattern from the end of the target pattern on a target pattern consisting of a pixel row including the defective pixels and a row or a plurality of pixel rows before and after. The defect pixel is calculated by calculating the correlation degree between the target pattern and the template while moving at least one pixel at a time along the direction, and using the central pixel of the portion of the target pattern for which the maximum correlation degree is calculated as a replacement pixel. The photoelectric conversion device according to claim 2, wherein the photoelectric conversion device is characterized in that. The noise detection unit sets the pixel value in the dark, which is a pixel value obtained in a state where no light is incident on the photoelectric conversion element that generates an image signal according to the amount of received light, at least for the first time and the first. The acquisition steps to be acquired in the second time, which is longer than the first time, and
Noise detection that detects the amount of noise in the dark by taking the difference between each pixel value in the dark acquired in the second time and each pixel value in the dark acquired in the first time. Steps and
A method for determining a defective pixel, wherein the determination unit has a determination step of determining a pixel having the noise amount whose difference is equal to or greater than a predetermined first threshold value as a defective pixel. An image forming apparatus that irradiates a document placed on a mounting table with light and receives the reflected light with a photoelectric conversion element to read the document.
The pixel value in the dark, which is a pixel value obtained when no light is incident on the photoelectric conversion element, is acquired at least in the first time and in the second time longer than the first time, respectively.
Noise detection that detects the amount of noise in the dark by taking the difference between each pixel value in the dark acquired in the second time and each pixel value in the dark acquired in the first time. Department and
A determination unit for determining a pixel having the noise amount whose difference is equal to or greater than a predetermined first threshold value as a defective pixel.
An image formation characterized by having a correction unit for correcting the pixel value of a pixel determined as the defective pixel among the pixel values of each pixel of the photoelectric conversion element to the pixel value of a normal pixel and outputting the pixel value. Device.

Images