JP7413917B2 - Image inspection device - Google Patents

Description

The present disclosure relates to an image inspection apparatus that inspects an image formed on a paper surface by an image forming apparatus, and a program that executes the image inspection.

2. Description of the Related Art Conventionally, image inspection apparatuses have been put into practical use that optically read an image formed on a paper surface by an image forming apparatus and inspect whether the image has been formed correctly. By inspecting an image with an image inspection device, it is possible to detect, for example, partial stains on the image.

When inspecting with an image inspection apparatus, it is necessary to correct the positional deviation between the original image and the image to be inspected formed on paper or the like to accurately align the two images.

Misalignment between the two images occurs when printing on paper and when scanning the printed image on the paper.

During printing, positional deviations due to image shift, rotation, or magnification occur due to the image writing start position, the inclination of the paper, and the shrinkage of the paper during fixing. During scanning, positional deviations due to image shift, rotation, or magnification occur due to variations in the image reading start position, paper inclination, and scan speed.

Among these factors, shrinkage of paper during fixing and fluctuations in scanning speed vary greatly depending on the device model and paper type. Therefore, there is almost no problem when comparing image data of two sheets scanned by the same device, but when comparing the original image data before printing and the image data obtained by scanning the image printed on paper. In some cases, a relatively large positional shift could occur.

In this respect, there is a possibility that variations in pixel values will become large, especially in areas where the density changes rapidly, such as in edge areas.

Therefore, in Japanese Patent Laid-Open No. 2018-155736, an edge region is extracted from the original image, and the threshold value of the extracted edge region is relaxed to prevent false detection in the region.

However, even if the threshold value for the edge region is relaxed, if the expected amount of deviation is large, it is necessary to widen the edge region accordingly to prevent false detection. If the expected amount of deviation is large, there is a possibility that the area that cannot be inspected will increase accordingly.

In particular, when attempting to inspect variable printing (such as forms) or items with many characters, if the edge area is wide, the area that can actually be inspected is limited. Therefore, in order to reduce the amount of deviation, it is necessary to perform alignment at high resolution, or to cope with nonlinear deviations such as distortion, it is necessary to divide the image and perform alignment for each divided area.

In Japanese Patent Application Laid-Open No. 2013-123812, since alignment accuracy is falsely detected in edge areas as low, it is determined whether an edge exists, and if it is determined that an edge exists, high-precision alignment is performed. A method has been proposed. In this respect, the area for which highly accurate positioning is to be performed is performed with high resolution, and the amount of positional deviation is determined and positioning is performed. Although the area is limited in this publication, increasing the alignment resolution or dividing the area to be aligned increases calculation costs and makes real-time processing difficult. Furthermore, when an image has only edges, it is necessary to align the entire surface at high resolution, making it difficult to process in real time.

Japanese Patent Application Publication No. 2018-155736 Japanese Patent Application Publication No. 2013-123812

The present disclosure has been made to solve the above problems, and an object of the present disclosure is to provide an image inspection apparatus that can perform highly accurate positioning using a simple method.

An image inspection apparatus according to one aspect includes an image acquisition unit that acquires a read image obtained by reading a printed matter, a reference image acquisition unit that acquires a reference image that serves as an inspection reference for the printed matter, and a read image of a first resolution and a reference image. A first positional deviation amount between the read image of the first resolution and the reference image is calculated based on the degree of similarity between the read image in the shifted state and the reference image, and the first positional deviation amount is calculated based on the calculation result. an alignment calculation unit that calculates a second positional deviation amount between a read image of a second resolution higher than the reference image and a reference image; an alignment correction section that aligns the read image after conversion and the reference image; and a difference inspection section that inspects based on a difference image between the read image of the first resolution and the reference image after alignment by the alignment section. Be prepared.

Preferably, the alignment calculation unit calculates a difference value between the maximum similarity among the similarities between the read image and the reference image and the similarity around it, and calculates the second positional deviation amount based on the calculation result. calculate.

Preferably, the alignment calculation unit has a plurality of threshold values to which the difference value is compared, and uses the first positional deviation amount as is based on a comparison result between the difference value and one of the plurality of threshold values.

Preferably, the alignment correction section shifts and aligns either the read image or the reference image.

Preferably, the alignment calculation unit converts the first resolution to the second resolution by bilinear interpolation or bicubic interpolation.

Preferably, the alignment calculation unit performs overall alignment of the read image of the first resolution and the reference image, divides the entire read image and the reference image after the overall alignment into a plurality of regions, and divides the read image and the reference image. For each region, the read image of the first resolution and the reference image are shifted within the search range, and the read image of the first resolution and the reference image are determined based on the similarity between the shifted read image and the reference image. A first positional deviation amount is calculated, and a second positional deviation amount between the reference image and a read image of a second resolution higher than the first resolution is calculated based on the calculation result.

Preferably, the image forming apparatus further includes an edge region generation unit that generates an edge region by extracting a region with large density variation from the reference image.

Preferably, the inspection unit adjusts the threshold value in the edge region.
Preferably, the edge region generation unit adjusts the width of the edge region based on the amount of distortion of the read image.

These and other objects, features, aspects and advantages of the present invention will become apparent from the following detailed description of the invention, taken in conjunction with the accompanying drawings.

1 is a diagram illustrating the appearance of an image inspection apparatus 1 according to a first embodiment. FIG. 1 is a diagram illustrating the configuration of an image inspection apparatus 1 according to a first embodiment. 1 is a block diagram showing the main hardware configuration of an image inspection apparatus 1 according to a first embodiment. FIG. FIG. 3 is a diagram illustrating a conventional inspection method as a comparative example. FIG. 3 is a diagram illustrating a positioning method for each divided region according to the first embodiment. FIG. 3 is a diagram illustrating a calculation result of ZNCC according to the first embodiment. FIG. 6 is a diagram illustrating calculation of the direction of the amount of deviation according to the first embodiment. FIG. 3 is a diagram illustrating a model of ZNCC shift amount according to the first embodiment. FIG. 3 is a conceptual diagram of shift amount interpolation according to the first embodiment. FIG. 2 is a diagram illustrating an inspection flow of the image inspection apparatus according to the first embodiment. FIG. 3 is a diagram illustrating a simulation example according to the first embodiment. FIG. 3 is a diagram illustrating brightness values according to coordinate values as a simulation example according to the first embodiment. FIG. 3 is a diagram illustrating determination of dirt in an image having an edge area as a simulation example according to the first embodiment. 7 is a diagram illustrating an edge region generated by an edge region generation unit 9 according to a third modification of the first embodiment. FIG. FIG. 7 is a diagram illustrating a ZNCC model according to the edge amount according to the second embodiment. FIG. 7 is a diagram illustrating a case where an average value of differences is calculated according to the second embodiment.

Hereinafter, embodiments of the technical idea according to the present disclosure will be described with reference to the drawings. In the following description, the same parts are given the same reference numerals. Their names and functions are also the same. Therefore, detailed descriptions thereof will not be repeated.

In the following embodiments, the image inspection device includes, for example, an MFP, a printer, a copying machine, or a facsimile.
(Embodiment 1)
[1. Configuration of image inspection device 1]
FIG. 1 is a diagram illustrating the appearance of an image inspection apparatus 1 according to the first embodiment. Referring to FIG. 1, an image inspection apparatus 1 is shown as a color printer. Although the image inspection apparatus 1 as a color printer will be described below, the image inspection apparatus 1 is not limited to a color printer. For example, the image inspection apparatus 1 may be a monochrome printer, or a multifunction device (so-called MFP (Multi Functional Peripheral)) that includes a monochrome printer or a color printer and a facsimile.

The image inspection apparatus 1 includes a scanner 20 as an image reading section and a printer 25 as an image forming section. The scanner 20 is provided with an ADF (Auto Document Feeder) 24. The printer 25 is provided with a tray 37 that stores paper. The image inspection apparatus 1 includes a display panel 105 for displaying various setting operations and information regarding the image inspection apparatus 1. The display panel 105 is provided on the front side of the image inspection apparatus 1.

FIG. 2 is a diagram illustrating the configuration of the image inspection apparatus 1 according to the first embodiment. Referring to FIG. 2, a scanner 20 is provided at the top of the image inspection apparatus 1. The scanner 20 includes a cover 21, a paper table 22, a tray 23, and an ADF 24. One end of the cover 21 is fixed to the paper table 22, and the cover 21 is configured to be openable and closable using the one end as a fulcrum. The operator can set the document on the paper table 22 by opening the cover 21. When the image inspection apparatus 1 receives a scan instruction with a document set on the paper table 22, it starts scanning the document set on the paper table 22. Further, when the image inspection apparatus 1 receives a scan instruction with documents set on the tray 23, the ADF 24 automatically reads the documents one by one.

The printer 25 includes image creation units 90Y, 90M, 90C, and 90K, an IDC (Image Density Control) sensor 19, a transfer belt 30, a primary transfer roller 31, a transfer drive device 32, a secondary transfer roller 33, It includes trays 37A to 37C, a driven roller 38, a driving roller 39, a timing roller 40, a cleaning unit 43, a fixing device 60, a reading device 2, and a control device 101.

The image forming units 90Y, 90M, 90C, and 90K are arranged in order along the transfer belt 30. The image forming section 90Y receives toner from the toner bottle 15Y and forms a yellow (Y) toner image. The image forming section 90M receives toner from the toner bottle 15M and forms a magenta (M) toner image. The image forming section 90C receives toner from the toner bottle 15C and forms a cyan (C) toner image. The image forming section 90K receives toner from the toner bottle 15K and forms a black (BK) toner image. The image forming units 90Y, 90M, 90C, and 90K are arranged along the transfer belt 30 in the order of the rotation direction of the transfer belt 30, respectively. The image creation units 90Y, 90M, 90C, and 90K each include a rotatable photoreceptor 10, a charging device 11, an exposure device 13, a developing device 14, a cleaning unit 17, and a toner sensor 18. Be prepared. After the image forming units 90Y, 90M, 90C, and 90K operate as described above, the transfer drive unit 32 transfers a yellow (Y) toner image, a magenta (M) toner image, and a cyan (C) toner image. The toner image and the black (BK) toner image are transferred from the photoreceptor 10 to the transfer belt 30 in a superimposed manner. As a result, a color toner image is formed on the transfer belt 30.

The IDC sensor 19 detects the density of the toner image formed on the transfer belt 30. Typically, the IDC sensor 19 is a light intensity sensor made of a reflective photosensor, and detects the intensity of reflected light from the surface of the transfer belt 30. The transfer belt 30 is stretched around a driven roller 38 and a driving roller 39. Drive roller 39 is connected to a motor (not shown). By controlling the motor, the drive roller 39 rotates. The transfer belt 30 and the driven roller 38 rotate in conjunction with the drive roller 39. As a result, the toner image on the transfer belt 30 is sent to the secondary transfer roller 33. Paper sheets of different sizes are set in each of the trays 37A to 37C. Paper is an example of a recording medium. Paper is conveyed one by one from one of the trays 37A to 37C to the timing roller 40 via a paper feeding mechanism. Then, the timing roller 40 sends it to the secondary transfer roller 33 along a conveyance path 41 . The transfer voltage applied to the secondary transfer roller 33 is controlled in accordance with the timing at which the paper is sent out.

The secondary transfer roller 33 applies a transfer voltage having a polarity opposite to the charged polarity of the toner image to the paper being transported. As a result, the toner image is attracted from the transfer belt 30 to the secondary transfer roller 33, and the toner image on the transfer belt 30 is transferred. The timing of transporting the paper to the secondary transfer roller 33 is controlled by a timing roller 40 in accordance with the position of the toner image on the transfer belt 30. As a result, the toner image on the transfer belt 30 is transferred to an appropriate position on the paper. The fixing device 60 presses and heats the paper passing through the fixing device 60. This fixes the toner image on the paper. Thereafter, the paper is discharged onto the tray 49. The cleaning unit 43 collects toner remaining on the surface of the transfer belt 30 after the toner image is transferred from the transfer belt 30 to the paper. The collected toner is transported by a transport screw (not shown) and stored in a waste toner container (not shown).

The reading device 2 includes a line image sensor that reads images formed on paper.
[2. Hardware configuration]
FIG. 3 is a block diagram showing the main hardware configuration of the image inspection apparatus 1 according to the first embodiment. An example of the hardware configuration of the image inspection apparatus 1 will be described with reference to FIG. 3.

In addition to a scanner 20 and a printer 25, the image inspection apparatus 1 includes a control device 101, a ROM (Read Only Memory) 102, a RAM (Random Access Memory) 103, a network interface 104, a display panel 105, and a microphone 106. , and a storage device 110. Control device 101 is configured, for example, by at least one integrated circuit. The integrated circuit includes, for example, at least one CPU (Central Processing Unit), at least one ASIC (Application Specific Integrated Circuit), at least one FPGA (Field Programmable Gate Array), or a combination thereof. The control device 101 controls the operation of the image inspection device 1 by executing various programs such as a program 112 for adjusting control parameters of the image inspection device 1 . The control device 101 reads the program 112 from the storage device 110 to the RAM 103 based on receiving the instruction to execute the program 112 . The RAM 103 functions as a working memory and temporarily stores various data necessary for executing the program 112. An antenna (not shown), a wireless module, and the like are connected to the network interface 104. The image inspection apparatus 1 exchanges data with external communication equipment via an antenna and a wireless module. External communication devices include, for example, mobile communication terminals such as smartphones, servers, and the like. The image inspection apparatus 1 may be configured to be able to download the program 112 from a server via an antenna.

The display panel 105 includes a display and a touch panel. The display and the touch panel are stacked on top of each other, and receive operations on the image inspection apparatus 1 through touch operations. Further, the display panel 105 can not only accept setting operations but also provide various information to the user.

The storage device 110 is, for example, a hard disk, SSD (Solid State Drive), or other storage device. The storage device 110 may be either a built-in type or an external type. The storage device 110 stores programs 112 and the like according to the embodiment. However, the storage location of the program 112 is not limited to the storage device 110, and may be stored in the storage area of the control device 101 (eg, cache), ROM 102, RAM 103, external equipment (eg, server), etc. The program 112 may be provided not as a standalone program but as a part of any program. In this case, the control processing according to the embodiment is realized in cooperation with an arbitrary program. Even if the program does not include some of these modules, it does not depart from the spirit of the program 112 according to the embodiment. Furthermore, some or all of the functionality provided by program 112 may be realized by dedicated hardware. Furthermore, the image inspection apparatus 1 may be configured as a so-called cloud service in which at least one server executes part of the processing of the program 112. Microphone 106 accepts audio input.

The control device 101 includes a read image acquisition section 3 , a reference image acquisition section 4 , an alignment calculation section 5 , an alignment correction section 6 , an image inspection section 8 , and an edge region generation section 9 .

The reading device 2 acquires data of an output target image to be formed on a sheet of paper by an image forming section. The data of the target image (output target image) is also referred to as RIP (Raster Image Processor) data.

The alignment calculation unit 5 performs alignment calculation processing to calculate correction parameters (alignment parameters) for aligning the read image data obtained by the reading device 2 and the RIP data obtained by the reference image acquisition unit 4. conduct.

The alignment correction unit 6 uses the correction parameters calculated by the alignment calculation unit 5 to perform alignment correction processing to correct one or both of the RIP data and the read image data. The position alignment correction unit 6 performs magnification processing, shift processing, and rotation processing. The scaling process is performed to correct a size error between the RIP data and the read image data. The shift process is performed to correct an error in image position between the RIP data and the read image data. The rotation process is performed to correct an error in the image rotation state between the RIP data and the read image data.

The image inspection section 8 determines defects in the read image by detecting a difference using the data corrected by the alignment correction section 6. The results determined by the image inspection unit 8 are displayed on a display panel 105 connected to the image inspection apparatus 1. Also, the paper conveyance in the conveyance unit is controlled based on the result determined by the image inspection unit 8, and the paper on which the defective image is formed is sent to a paper output tray different from the normal paper output tray. You can also do this.

The edge region generation unit 9 generates edge regions by extracting regions with large density fluctuations from the RIP data. The image inspection unit 8 relaxes the threshold value for stain determination in the generated edge region.

FIG. 4 is a diagram illustrating a conventional inspection method as a comparative example. As shown in FIG. 4(A), a read image that is an inspection image is shown. FIG. 4B shows a reference image that is a RIP image. FIG. 4C shows a difference image between the inspection image and the reference image. Black indicates a small difference, and white indicates a large difference. Large differences occur around edges, such as around characters. In order to avoid false detections around edges, it is necessary to relax the threshold. If the threshold value is relaxed to avoid false detection around edges, dirt around characters cannot be detected. Therefore, in the embodiment, by dividing the region and aligning each divided region, it is possible to cope with partial distortion.

In the first embodiment, since the maximum amount of distortion varies depending on the paper type and paper size, the edge region generation unit 8 outputs a chart for measuring the distortion from the selected tray and calculates the maximum amount of distortion. In this regard, a chart with regular repeating images, such as a grid, is output and read. It is also possible to search for images that are deviated from the intended position by measuring the positions of repeated images, and calculate the amount of distortion from the maximum positional deviation amount as a result of the search. If the amount of distortion exceeds the search range that can be processed in real time, either increase the amount of edges that relax the threshold and expand the search range to the one that can be processed in real time, or expand the search range and reduce productivity in order to be able to process in real time. Alternatively, the settings may be made from the panel.

A method of positioning each divided area according to the first embodiment will be described. FIG. 5 is a diagram illustrating a positioning method for each divided region according to the first embodiment.

As shown in FIG. 5(A), one of the two divided reference images and inspection image is cut out by an amount of distortion. In this example, a case is shown where only the periphery of the reference image is cut out by an assumed amount of distortion. As an example, a case is shown in which the maximum amount of distortion is 0.38 mm (6dots@200dpi), and the surrounding area is cut out at 6dots@200dpi. Note that the resolution of the inspection image and reference image is set to 200@dpi. In this example, a case will be described in which a deviation amount of 800 dpi is calculated.

As shown in FIG. 5B, while shifting the cut reference image, the degree of similarity between the area corresponding to the image and the test image is calculated. As an example, the degree of similarity is calculated using ZNCC (Zero-mean Normalized Cross-Correlation).

FIG. 6 is a diagram illustrating the calculation results of ZNCC according to the first embodiment. As shown in FIG. 6, the maximum coordinates of the ZNCC with the highest degree of similarity are x, y=9,6. Furthermore, in order to calculate in which direction the position should be shifted from here, the direction of the amount of shift is determined using the result of ZNCCC at the maximum coordinate and the ZNCC results at eight surrounding locations.

FIG. 7 is a diagram illustrating calculation of the direction of the amount of deviation according to the first embodiment. As shown in FIG. 7(A), calculations are made for the upper left, lower left, upper right, and lower right directions. FIGS. 7(B), (C), (D), and (E) are diagrams illustrating each of the four locations: upper left, upper right, lower left, and lower right. In this example, the total of each of the four locations is calculated. It is assumed that the deviation is in the maximum direction among the four totals. In this example, the total value at the bottom right is the maximum value, so it is shifted toward the bottom right. Next, after calculating the direction of deviation, the amount of deviation with respect to the direction of deviation is calculated. In order to calculate the amount of deviation in the x direction, the difference between the maximum ZNCC result at the center and the ZNCC result one position to the right is calculated.

In the first embodiment, the amount of deviation is calculated based on the comparison between the difference and the threshold value. In this example, a case will be described in which two threshold values TH11 and TH2 are provided in order to calculate the amount of shift in resolution of 4 times. The relationship between threshold value TH1 and threshold value TH2 is set such that threshold value TH1>threshold value TH2. The number of threshold values may be changed depending on the resolution of the image to be shifted.

FIG. 8 is a diagram illustrating a model of the ZNCC shift amount according to the first embodiment. FIG. 8A shows a case (Model 1) in which the ZNCC result obtained at 200 dpi matches the maximum position of ZNCC obtained at 800 dpi. In the case of this model 1, as shown in the figure, there is a large difference between the result of the maximum value ZNCC and the result of the adjacent ZNCC. Therefore, if the difference between the maximum ZNCC result and the neighboring ZNCC result is greater than the threshold TH1, it is determined that the ZNCC value obtained at 200 dpi is classified into model 1. The value obtained by multiplying the positional deviation amount obtained at 200dpi by 4 is set as the positional deviation amount of 800dpi.

FIG. 8B shows a case (Model 2) in which the ZNCC result obtained at 200 dpi is shifted by 1 dot to the left of the maximum position of ZNCC obtained at 800 dpi. In the case of this model 2, the difference between the result of the maximum ZNCC and the result of the adjacent ZNCC is smaller than that of the model 1, as shown in the figure. Therefore, if the difference between the maximum ZNCC result and the adjacent ZNCC result is between the threshold TH1 and the threshold TH2, it is determined that the ZNCC result obtained at 200 dpi is classified into model 2. Then, the value obtained by multiplying the positional deviation amount obtained at 200 dpi by 4 and ±1 is set as the positional deviation amount at 800 dpi. Note that if it is shifted to the right or down, set it to "+", and if it is shifted to the left or above, set it to "-".

FIG. 8C shows a case (Model 3) where the ZNCC result obtained at 200 dpi is shifted 2 dots to the left from the maximum position of ZNCC obtained at 800 dpi. In the case of this model 3, as shown in the figure, the difference between the result of the maximum ZNCC and the result of the adjacent ZNCC is even smaller. Therefore, if the difference between the maximum ZNCC result and the adjacent ZNCC result is smaller than the threshold TH2, it is determined that the ZNCC result obtained at 200 dpi is classified into model 3. Then, the value obtained by multiplying the positional deviation amount obtained at 200 dpi by 4 and ±2 is set as the positional deviation amount at 800 dpi. Note that if it is shifted to the right or down, set it to "+", and if it is shifted to the left or above, set it to "-".

In addition, if the ZNCC result obtained at 200 dpi is shifted 3 dots to the left of the maximum position of ZNCC obtained at 800 dpi, it is the same as model 2, which is shifted in the opposite direction, and the difference obtained in advance is the same as model 2, which is shifted in the opposite direction. The predicted direction is the opposite direction.

FIG. 9 is a conceptual diagram of shift amount interpolation according to the first embodiment. FIG. 9 shows the amount of shift when enlarging an image with a resolution of 200 dpi to 800 dpi, which is four times the resolution.

If the difference between the maximum value ZNCC result and the adjacent ZNCC result is larger than the threshold TH1, the value obtained by multiplying the maximum coordinate -6 value by 4 is set as the deviation amount for the 800 dpi image. Also, if the difference between the maximum ZNCC result and the adjacent ZNCC result is between the threshold TH1 and the threshold TH2, the maximum coordinate - 6 value multiplied by 4 plus ±1 is calculated at 800 dpi. This is the amount of deviation from the image. Also, if the difference between the maximum value ZNCC result and the adjacent ZNCC result is smaller than the threshold TH2, the value obtained by multiplying the value of the maximum coordinate -6 by 4 and ±2 is used as the deviation amount for the 800 dpi image. do. Note that when the ZNCC result is shifted by 3 dots to the left, it is the same as when it is shifted by 1 dot to the right from the ZNCC result immediately to the left.

In the case of this example, the threshold value TH1 is set to 0.04, and the threshold value TH2 is set to 0.02. The difference between the maximum value ZNCC and the adjacent ZNCC on the right side is 0.044, and the difference between the ZNCC on the lower side is 0.023. Therefore, based on the above method, the amount of shift for an 800 dpi image is set to x, y=(9-6)×4=12, (6-6)×4+1=1. Next, in order to shift the 200 dpi reference image by x, y = 12, 1@800 dpi, bilinear interpolation is performed to expand the resolution to 800 dpi. Next, after enlarging the image to 800 dpi, shift the image by x, y = 12, 1@800 dpi, and perform bilinear interpolation to reduce the resolution to 200 dpi.

Next, the difference between the reference image, whose resolution has been reduced to 200 dpi, and the inspection image is taken, and the threshold value of the edge area is relaxed and threshold processing is performed, and as a result, it is determined that there is dirt in areas where the threshold value is exceeded.

FIG. 10 is a diagram illustrating an inspection flow of the image inspection apparatus according to the first embodiment. As shown in FIG. 10, the image inspection apparatus 1 acquires a RIP image (step S2). The reference image acquisition unit 4 acquires RIP data. Next, the image inspection apparatus 1 converts the resolution of the RIP image (step S4). The alignment calculation unit 5 converts the resolution. This resolution conversion is performed to match the resolution of the RIP data with the resolution of the read image data. For example, the alignment calculation unit 5 converts the resolution of the RIP data from 800 dpi to 200 dpi. Next, the image inspection device 1 performs color conversion (step S6). The alignment calculation unit 5 calculates the primary colors of the additive color mixing method from the data of the color components of C (cyan), M (magenta), Y (yellow), and K (black), which are the primary color components of the subtractive color mixing method for printing. It is converted into color component data of R (red), G (green), and B (blue). Next, the image inspection apparatus 1 stores the information (step S8). The alignment calculation unit 5 stores the converted RGB data in the RAM 103 as model data. Note that if the image data includes a variable area (an area where image content differs for each copy), information in the variable area is removed from the model data. Next, the image inspection apparatus 1 acquires an inspection image (step S10). The read image acquisition unit 3 acquires a read image, which is a test image, from the reading device 2 . The read image data acquired by the reading device 2 is RGB data with a resolution of 200 dpi. Next, the image inspection apparatus 1 calculates correction parameters for the model data and the inspection image (step S12). The alignment calculation unit 5 compares the model data stored in the RAM 103 with the acquired read image data and calculates correction parameters. The correction parameters consist of image shift, rotation and magnification.

Next, the image inspection apparatus 1 performs overall alignment (step S14). The alignment correction unit 6 executes overall alignment using the correction parameters.

Next, the image inspection apparatus 1 executes division processing (step S16). After the overall alignment, the alignment calculation unit 5 divides the inspection image and the reference image into a plurality of regions. In particular, when aligning with a RIP image, the image is deformed non-linearly with respect to the RIP image due to misalignment of printing, shrinkage of paper, and the like. Further, the inspection image (scanned image) is partially distorted due to the behavior of the paper during scanning.

Next, the image inspection apparatus 1 calculates the amount of deviation (step S18). The alignment calculation unit 5 calculates the amount of deviation according to the method described in FIGS. 5 to 9. Next, the image inspection apparatus 1 performs resolution conversion (step S20). The alignment calculation unit 5 enlarges the resolution of the 200 dpi reference image to 800 dpi by bilinear interpolation. Next, the image inspection apparatus 1 executes alignment (step S22). After enlarging the image to 800 dpi, the alignment correction unit 6 shifts the image by the amount of deviation. Further, the alignment correction unit 6 performs bilinear interpolation to reduce the resolution to 200 dpi.

Next, the image inspection apparatus 1 inspects (step S24). The image inspection unit 8 calculates the difference between the reference image whose resolution has been reduced to 200 dpi and the inspection image, relaxes the threshold of the edge area, performs threshold processing, and determines that there is dirt in areas exceeding the threshold. . The image inspection apparatus 1 then ends the inspection flow process (END).

FIG. 11 is a diagram illustrating a simulation example according to the first embodiment. FIG. 11A shows a state before the inspection image and the reference image are aligned. FIG. 11(B) shows the state after alignment of the inspection image and the reference image.

FIG. 12 is a diagram illustrating brightness values according to coordinate values as a simulation example according to the first embodiment. The luminance value is extracted from the rectangular area shown in FIG. As shown in FIG. 12, the brightness value of the inspection image, the brightness value of the reference image before alignment, and the brightness value of the reference image after alignment are shown. By aligning the positions, the conversion points of the inspection image and the reference image match, and inspection can be performed without false detection.

FIG. 13 is a diagram illustrating determination of dirt in an image having an edge region as a simulation example according to the first embodiment. As shown in FIG. 13, in this example, the amount of shift between the inspection image and the reference image is 2 dots. In this case, when the amount of shift is 0 dots, there is a case where it is erroneously detected that there is dirt in the difference image. The method according to the first embodiment allows highly accurate alignment with a simple method.

In this example, a method of performing resolution conversion and positioning for the reference image has been described, but a similar method may be adopted for the inspection image.

(Modification 1 of Embodiment 1)
In Embodiment 1 above, a case has been described in which the direction of the amount of deviation is calculated using the results of eight places around the maximum coordinate of ZNCC. The direction of the amount may also be calculated. For example, the ZNCC results above and below the coordinates of the maximum value are compared, and it is determined that the deviation is greater in the upper or lower direction. Specifically, the difference between the maximum ZNCC result and the upper and lower ZNCCs is taken to calculate the amount of deviation. Note that if the values are the same on the top and bottom, it is determined that there is no deviation in the vertical direction with respect to the coordinates of the maximum value.

Similarly, the ZNCC results on the left and right sides of the coordinates of the maximum value are compared, and it is determined that the deviation is in the direction that is larger between the left and right. Specifically, the difference between the maximum ZNCC result and the left and right ZNCCs is taken to calculate the amount of deviation. Note that if the values are the same on the left and right sides, it is determined that there is no deviation in the left and right direction with respect to the coordinates of the maximum value. The calculation of the amount of deviation after calculating the direction of deviation is the same as in the first embodiment described above. With this method, it is possible to further reduce calculation costs. By the method according to this embodiment, highly accurate alignment is possible with a simple method.

(Modification 2 of Embodiment 1)
In the first embodiment described above, a case has been described in which the direction of the shift amount is calculated using the results of eight locations around the maximum coordinate of ZNCC. Specifically, it was determined that the direction is the direction in which the total value becomes the maximum value, but for example, by using the coordinates of the maximum value of the ZNCC results and the ZNCC results of the surrounding coordinates, find the center of gravity, Calculate the direction of deviation. Alternatively, the amount of deviation may be calculated using a threshold value obtained by subtracting the position coordinates of the center of gravity and the maximum position coordinate of the ZNCC.

Specifically, if the value obtained by subtracting the position coordinates of the center of gravity and the maximum position coordinate of ZNCC is smaller than the threshold TH3, the value obtained by multiplying the value of the maximum coordinate - 6 by 4 is set as the amount of deviation for the 800 dpi image. . Also, if the value obtained by subtracting the position coordinates of the center of gravity and the maximum position coordinate of ZNCC is smaller than the threshold TH4, the value obtained by multiplying the value of the maximum coordinate - 6 by 4 and +1 is determined as the amount of deviation for the 800 dpi image. do. Also, if the value obtained by subtracting the position coordinates of the center of gravity and the maximum position coordinate of ZNCC is larger than the threshold TH4, the value obtained by multiplying the value of the maximum coordinate -6 by 4 and +2 is calculated as the amount of deviation for the 800 dpi image. do. Here, the relationship between threshold value TH3 and threshold value TH4 is threshold value TH3<threshold value TH4. The number of thresholds depends on the resolution of the image to be shifted. If it is off to the right or down, mark it as “+”; if it is off to the left or on top, mark it as “-”.

With this method, it is possible to calculate the positional coordinates of the center of gravity and calculate the amount of deviation with high accuracy. With this method, highly accurate alignment is possible using a simple method.

(Variation 3 of Embodiment 1)
FIG. 14 is a diagram illustrating an edge region generated by the edge region generation unit 9 according to the third modification of the first embodiment. As shown in FIG. 14, the right and left sides of the edge region are shown. Regarding the edge region, it is possible to suppress false determinations by relaxing the threshold value. The edge region generation unit 9 may adjust the width of the edge region to be extracted based on the amount of distortion of the read image. Specifically, when the amount of distortion is measured and exceeds a threshold value, the width of the edge region may be narrowed, and when it is equal to or less than the threshold value, the width of the edge region may be widened. When the width of the edge region is narrowed, the search range is widened, and when the width of the edge region is widened, the search range is narrowed. Therefore, the width of the edge region may be adjusted by selecting whether to reduce the productivity of the machine or give priority to the productivity of the machine. This method enables inspection based on highly accurate alignment using a simple method.

(Embodiment 2)
FIG. 15 is a diagram illustrating a ZNCC model according to the edge amount according to the second embodiment. As shown in FIG. 15A, in the case of an image with many edges, that is, with large density fluctuations, the difference increases when the positions of the reference image and the inspection image shift. Therefore, the difference between the maximum ZNCC and the result of the ZNCC of the coordinate next to the coordinate with the maximum ZNCC also becomes large.

As shown in FIG. 15(B), if the image has few edges, that is, small density fluctuations, the difference will be small even if the positions of the reference image and the inspection image are shifted. Therefore, the difference between the maximum ZNCC and the result of the ZNCC of the coordinate next to the coordinate with the maximum ZNCC becomes small. Therefore, the trends are different.

The tendency of the calculated cross-correlation differs not only depending on the type of image depending on the amount of edges, but also when the edges are clearly visible and when they are blurred, depending on the sampling phase during reading. Furthermore, the tendency of the calculated cross-correlation also differs when an image read during proof printing is used instead of an RIP image as the reference image. Therefore, for example, in order to be able to classify using the same threshold, take the difference between the maximum ZNCC and the results of the ZNCCs at eight locations around the coordinates of the maximum ZNCC, and calculate the difference in the direction in which the maximum ZNCC result deviates from the coordinates of the maximum ZNCC. The difference with the ZNCC result may be normalized by dividing the difference by the average value of the difference, and a high-resolution shift amount may be obtained by performing threshold processing.

FIG. 16 is a diagram illustrating the case of calculating the average value of differences according to the second embodiment. FIG. 16(A) shows the same ZNCC results as described in FIG. 6. The center is the maximum value. FIG. 16(B) shows the difference from the center (maximum value). In FIG. 16(C), the average value of the differences is calculated. Then, the difference from the center is divided by the average value of the differences. As described above, the direction of deviation is calculated by calculating the sum of the surrounding four pixels including the maximum value. In this example, the total value at the bottom right is the maximum value, so it is shifted toward the bottom right.

In this example, the threshold value TH1 is assumed to be 0.65, and the threshold value TH2 is assumed to be 0.2. The difference with the right pixel is 0.75, and the difference with the lower pixel is 0.4. The amount of shift for an 800 dpi image is x=(9-6)×4=12, y=(6-6)×4+1=1.

Using this method, the difference between the maximum ZNCC and the results of the ZNCCs at 8 locations around the coordinates of the maximum ZNCC is calculated, and the difference between the maximum ZNCC result and the ZNCC result in the deviating direction is calculated as the average value of the differences. It is possible to normalize by dividing by , and to obtain a high-resolution shift amount by performing threshold processing. That is, since calculation can be performed using a common threshold value, calculation cost can be suppressed, and highly accurate positioning can be performed using a simple method.

The embodiments disclosed this time should be considered to be illustrative in all respects and not restrictive. The scope of the present invention is indicated by the claims rather than the above description, and it is intended that all changes within the meaning and range equivalent to the claims are included.

1 image inspection device, 2 reading device, 3 read image acquisition unit, 4 reference image acquisition unit, 5 alignment calculation unit, 6 alignment correction unit, 8 image inspection unit, 9 edge area correction unit, 10 photoreceptor, 11 charging Device, 13 Exposure device, 14 Developing device, 15C, 15K, 15M, 15Y Toner bottle, 25 Printer, 30 Transfer belt, 31 Primary transfer roller, 35 Paper feeding mechanism, 38 Followed roller, 39 Drive roller, 40 Timing roller, 41 transport path, 60 fixing device, 101 control device, 102 ROM, 103 RAM, 104 network interface, 105 display panel, 106 microphone, 110 storage device.

Claims (9)

an image acquisition unit that acquires a read image of the printed matter;
a reference image acquisition unit that acquires a reference image that serves as an inspection reference for the printed matter;
The read image at the first resolution and the reference image are shifted within a search range, and the read image at the first resolution and the reference image are determined based on the degree of similarity between the shifted read image and the reference image. an alignment calculation unit that calculates a first positional deviation amount between the reference image and the read image at a second resolution higher than the first resolution based on the calculation result; and,
an alignment correction unit that aligns the read image after conversion from the first resolution to the second resolution and the reference image based on the calculated second positional deviation amount;
An image inspection apparatus comprising: a difference inspection section that inspects based on a difference image between the read image of the first resolution and the reference image after alignment by the alignment section. The alignment calculation unit calculates a difference value between the maximum similarity among the similarities between the read image and the reference image and the similarity around it, and calculates the second positional deviation amount based on the calculation result. The image inspection apparatus according to claim 1, which calculates . The alignment calculation unit has a plurality of threshold values to be compared with the difference value,
The image inspection apparatus according to claim 1, wherein the first positional deviation amount is used as it is based on a comparison result between the difference value and one of the plurality of threshold values. The image inspection apparatus according to claim 1, wherein the alignment correction section aligns one of the read image or the reference image by shifting it. The image inspection apparatus according to claim 1, wherein the alignment calculation unit converts the first resolution to the second resolution by bilinear interpolation or bicubic interpolation. The alignment calculation unit includes:
Aligning the entire read image of the first resolution with the reference image, dividing the entire read image after the overall alignment and the reference image into a plurality of regions,
For each divided area, the read image at the first resolution and the reference image are shifted within the search range, and the image at the first resolution is determined based on the similarity between the shifted read image and the reference image. A first positional deviation amount between the read image and the reference image is calculated, and a second positional deviation amount between the read image at a second resolution higher than the first resolution and the reference image based on the calculation result. The image inspection apparatus according to claim 1, which calculates . The image inspection apparatus according to claim 1, further comprising an edge region generation unit that generates an edge region by extracting a region with a large density variation from the reference image. The image inspection apparatus according to claim 7, wherein the inspection section adjusts a threshold value in the edge area. The image inspection apparatus according to claim 7, wherein the edge area generation unit adjusts the width of the edge area based on the amount of distortion of the read image.

Images