JP6665903B2 - Feature image generation device, inspection device, feature image generation method, and feature image generation program - Google Patents

Description

  The present invention relates to a technique for aligning a reference image obtained by capturing an object serving as a reference and a comparison image obtained by capturing a comparison object, when comparing the two images.

  Conventionally, as for a method of automatically inspecting a printed matter printed by a printing machine on a printing machine for color tone quality, presence or absence of dirt, presence or absence of omission, an inspection image obtained by imaging the printed matter with a camera, There is a method of comparing reference images satisfying a certain reference.

  In such an inspection method, it is necessary to perform accurate alignment when comparing the reference image and the image to be inspected. For example, in rotary printing, printing is performed on a continuous sheet such as roll paper, but when an image to be inspected is obtained by imaging the printed continuous sheet during transport, the continuous sheet contracts in the transport direction or meanders in the width direction. It is necessary to perform positioning before the comparative inspection. Also, in sheet printing, printing is performed on a cut sheet such as a sheet, but when the cut sheet is printed and imaged to obtain an image to be inspected, the cut sheet is irregularly arranged in the vertical, horizontal, and rotational directions. Positioning is required for displacement.

  Patent Literatures 1 to 3 disclose techniques for performing position adjustment (sometimes referred to as position correction) of an image to be inspected so that the image can be compared with a reference image. These are techniques for subdividing an image to be inspected and a reference image into a plurality of blocks, calculating an image shift amount for each block, and performing overall position correction. Further, an image including a characteristic pattern (characteristic portion) is cut out for each block to generate a characteristic image, matching is performed between the image to be inspected and the reference image, and position correction is performed based on how much the characteristic portion is shifted. It is also known.

Japanese Patent No. 3916596 Japanese Patent No. 4163199 Japanese Patent No. 5061543

  On the other hand, technical advances in cameras and the like have made it possible to more precisely inspect printed matter using high-resolution images. However, the techniques disclosed in Patent Literatures 1-3 are suitable for low-resolution images, and may not be sufficiently corrected for high-resolution images. Also, the method of performing position correction using a characteristic portion has a problem that a high-resolution image cannot select a characteristic portion suitable for position correction within a block. This is thought to be due to the fact that the design becomes relatively large and abstract in the subdivided blocks due to the increase in resolution, and there are many blocks without designs due to the small block size. Can be

  The present invention has been made in view of such a problem, and one example of the problem is to appropriately generate a feature image used for matching for aligning a high-resolution reference image and a comparative image from the reference image. It is an object of the present invention to provide a characteristic image generation device or the like that can perform the above-mentioned operations.

  In order to solve the above-mentioned problem, the invention according to claim 1 includes an edge extraction unit that generates an edge image by extracting an edge from a reference image, and a first image that binarizes the edge image with a first gradation. A binarized image, a second binarized image obtained by binarizing the edge image with a second gradation lower than the first gradation, and the edge image with the first gradation and the second gradation. A third binarized image binarized at a third gray level between two gray levels, and a binarization unit that generates a third binarized image, and a horizontal straight line and a vertical straight line in the first binary image. An intersection acquisition unit that acquires intersection coordinates; and an extraction unit that extracts an image including the intersection coordinates from the first binarized image and the second binarized image and generates a first cutout image and a second cutout image, respectively. An image generation unit, wherein a difference pixel number between the first cutout image and the second cutout image is a predetermined condition. If the plus, characterized in that it and a feature image generation unit for generating feature images cut out an image including the intersection coordinates from said third binary image.

  The invention according to claim 2 is the feature image generation device according to claim 1, wherein a first horizontal image obtained by extracting a horizontal straight line from the first binary image, and the first binary image. A first vertical image extracting a vertical straight line from the first vertical image, and a first synthesizing unit for generating a first synthesized image; a second horizontal image extracting a horizontal straight line from the second binarized image; A second synthesizing unit for synthesizing a second vertical image obtained by extracting a vertical straight line from the binarized image to generate a second synthesized image; Acquiring the intersection coordinates based on the first synthesized image generated from the first synthesized image and the first synthesized image and the second binarized image generated from the first binarized image. From the second composite image generated from the image, the first cutout image and the second And generating an image Eject and.

  The invention according to claim 3 is the feature image generation device according to claim 2, wherein the first synthesizing unit extracts a horizontal straight line from the first binarized image and enlarges the horizontal straight line. Extracting a vertical straight line from the first horizontal image and the first binarized image, combining the first vertical image enlarged in the vertical direction, generating the first combined image, The combining unit extracts a horizontal straight line from the second binarized image, extracts the second horizontal image enlarged in the horizontal direction, and a vertical straight line from the second binarized image, and extracts a vertical straight line from the second binary image. The enlarged second vertical image is combined to generate the second combined image.

The invention according to claim 4 is the feature image generation device according to any one of claims 1 to 3, further comprising a division unit configured to divide the reference image into a plurality of blocks, The unit generates the feature image for the block including the intersection coordinates when the first cutout image and the second cutout image in which the difference pixel number satisfies the predetermined condition .

  The invention according to claim 5 is the feature image generating apparatus according to claim 4, wherein the reference image captures a part of the continuous sheet when the continuous sheet is conveyed in the longitudinal direction. When the image is an image, one feature image is generated for one block.

  The invention according to claim 6 is the feature image generation device according to claim 4, wherein the reference image is an image obtained by photographing a cut sheet obtained by cutting a continuous sheet into sheet units. It is characterized in that two feature images are generated for a block.

  The invention according to claim 7 is the feature image generation device according to any one of claims 1 to 6, wherein a verification area based on the intersection coordinates in the third binary image is The image processing apparatus further includes a determination unit that compares the feature image with the third binarized image while moving the feature image, and determines that the feature image is an inappropriate feature image when the feature image matches other than the intersection coordinates.

  According to an eighth aspect of the present invention, a position of an image to be inspected is corrected using the characteristic image created by the characteristic image generating apparatus according to any one of the first to seventh aspects, and the reference image and the image to be inspected are corrected. An inspection apparatus for performing a comparative inspection of an inspection image, comprising: identifying coordinates corresponding to the characteristic image in the image to be inspected; and calculating a shift amount that is a difference from the intersection coordinates acquired by the intersection acquisition unit. An amount calculator, a position corrector that corrects the position of the inspected image based on the deviation amount, and a comparator that compares the inspected image subjected to the position correction and the reference image. It is characterized by the following.

  The invention according to claim 9, wherein an edge extraction step of generating an edge image by extracting an edge from a reference image; a first binarized image obtained by binarizing the edge image with a first gradation; A second binarized image obtained by binarizing the image at a second gradation lower than the first gradation, and a second binarized image obtained by converting the edge image between the first gradation and the second gradation. A binarizing step of generating a third binarized image binarized by three gradations, and an intersection obtaining step of obtaining intersection coordinates of a horizontal straight line and a vertical straight line in the first binary image A cutout image generating step of cutting out an image including the intersection coordinates from the first binarized image and the second binarized image to generate a first cutout image and a second cutout image, respectively; When the difference pixel number between the cutout image and the second cutout image satisfies a predetermined condition, And wherein the image generating step from the 3 binary image to generate a feature image cut out an image including the intersection coordinates, characterized in that it comprises a.

  The invention according to claim 10 is a computer, comprising: an edge extraction unit that generates an edge image obtained by extracting an edge from a reference image; a first binarized image obtained by binarizing the edge image with a first gradation; A second binarized image obtained by binarizing the edge image with a second gradation lower than the first gradation, and converting the edge image between the first gradation and the second gradation. A third binarized image binarized at the third gradation, and an intersection acquisition unit for acquiring intersection coordinates of a horizontal straight line and a vertical straight line in the first binary image. A cutout image generation unit that cuts out an image including the intersection coordinates from the first binarized image and the second binarized image to generate a first cutout image and a second cutout image, respectively, and the first cutout When the difference pixel number between the image and the second cut image satisfies a predetermined condition, Feature image generation unit for generating feature images cut out an image including the intersection coordinates from the third binarized image, characterized in that to function as a.

  According to the present invention, a first clipped image obtained from an image obtained by binarizing an edge image obtained by extracting an edge from a reference image with a first gradation and an edge image is binarized by a second gradation When the difference between the second cut-out images obtained from the obtained images satisfies a predetermined condition, the same position is cut out from the image obtained by binarizing the edge image with the third gradation to obtain a feature image. If the gradation in the edge image of the image to be compared with the reference image is between the first gradation and the second gradation, the alignment of the reference image and the comparison image can be performed. That is, a feature image used for matching for aligning the high-resolution reference image and the comparative image can be appropriately generated from the reference image.

FIG. 2 is a diagram illustrating an example of a printing machine S according to the first embodiment. It is a block diagram showing an example of composition of inspection device T in a 1st embodiment. 6 is a flowchart illustrating an example of a continuous sheet template image creation process performed by the inspection device T according to the first embodiment. FIG. 3 is a diagram illustrating an example in which a reference image 200 according to the first embodiment is divided into a plurality of blocks. It is an example of the A image of the R component image. It is an example of the A image of the G component image. It is an example of the A image of the B component image. (A) is an example of a B image of an R component image, (B) is an example of a B image of a G component image, (C) is an example of a B image of a B component image, and (D) is an RGB image. It is an example of a B image of a component image. (A) is an example of a C image of an R component image, (B) is an example of a C image of a G component image, (C) is an example of a C image of a B component image, and (D) is an RGB image. It is an example of a C image of a component image. (A) is an example of a C1 image, (B) is an example of a C2 image, and (C) is an example of an image obtained by combining a C1 image and a C2 image. (A) is an example of a B1 image, (B) is an example of a B2 image, and (C) is an example of an image obtained by combining a B1 image and a B2 image. (A) is an example of a C3 image, (B) is an example of a B3 image, (C) is an example of a C 'image, and (D) is an example of a B' image. It is an example of a D image. 5 is a flowchart illustrating an example of a continuous sheet inspection process performed by the inspection device T according to the first embodiment. 6 is a flowchart illustrating an example of a continuous sheet shift amount calculation process performed by the inspection device T according to the first embodiment. 5 is a flowchart illustrating an example of a continuous sheet template presence block process performed by the inspection apparatus T according to the first embodiment. It is a flow chart which shows an example of sheet template image creation processing by inspection device T in a 2nd embodiment. It is a flow chart which shows an example of sheet inspection processing by inspection equipment T in a 2nd embodiment. It is a flow chart which shows an example of sheet and gap amount calculation processing by inspection device T in a 2nd embodiment. It is a flowchart which shows an example of the sheet | seat / template presence block process by the inspection apparatus T in 2nd Embodiment.

[1. First Embodiment]
Hereinafter, a first embodiment of the present invention will be described with reference to the drawings. The embodiment described below is an embodiment in which the present invention is applied to an inspection device installed in an offset rotary printing press.

[1.1. Configuration of offset rotary printing press and inspection device]
The configuration of the offset rotary printing press S (hereinafter, referred to as “printing press S”) and the inspection device T according to the present embodiment will be described with reference to FIGS. 1 and 2.

[1.1.1. Configuration of Printing Machine S]
First, the configuration of the printing machine S will be described with reference to FIG. The printing machine S is a multi-color printing machine, and is provided with printing units U1 to U4 for each ink color (C (cyan), M (magenta), Y (yellow), and K (black)). Since the printing machine S is a two-sided printing machine, each printing unit U1-U4 is provided with a blanket cylinder (upper cylinder, lower cylinder) so as to sandwich the paper transport path, and a plate cylinder (upper cylinder) is provided for each blanket cylinder. , Lower drum) and an ink supply device (upper drum side, lower drum side). The plate cylinder is provided with a plate of a picture for one sheet, and the blanket cylinder is provided with ink for a picture of two sheets. When the plate cylinder makes two rotations, the blanket cylinder makes one rotation. (So-called "double-blanket blanket cylinder"). That is, a picture for two cut sheets is printed while the blanket cylinder makes one rotation. Note that, in the present embodiment, a case is described in which the print target is a band-shaped roll paper (an example of a “continuous sheet”), but the present invention is also applicable to a continuous sheet other than the roll paper. The material of the continuous sheet may be cloth or vinyl as long as it can be printed. A sheet obtained by cutting a continuous sheet (for example, sheet) is referred to as a cut sheet.

  A dryer U5 is provided downstream of the printing units U1-U4. In addition, downstream of the dryer U5, an inspection target area (including an image portion and a non-image portion forming a pattern) including a pattern (symbol, graphic, photograph, pattern, and the like) printed on roll paper as a printing material is included. For example, a front camera 31A and a front illumination 32A, and a back camera 31B and a rear illumination 32B are provided in order to capture an image of a newspaper page. The front camera 31A and the back camera 31B (hereinafter, may be collectively referred to as “camera 31”) are line sensor cameras that can capture a line image of 4096 pixels every 25 μsec, for example. A camera signal obtained by imaging the printed roll paper is output. The front illumination 32A and the back illumination 32B (hereinafter sometimes collectively referred to as “illumination 32”) irradiate sufficient light for the front camera 31A and the rear camera 31B to capture a high-resolution image, respectively.

  A folding machine U6 for folding roll paper on which a picture is printed is provided further downstream of the portion where the camera 31 and the lighting 32 are installed.

[1.1.2. Configuration of inspection device T]
Next, the configuration of the inspection device T will be described with reference to FIG. The inspection apparatus T includes a system controller 11, a pipeline image processing unit 12, a parallel image processing unit 13, a front camera branch board 14, a back camera branch board 15, an encoder branch board 16, a HUB 17, a camera 31, a lighting 32, and an encoder 33. It is comprised including. The inspection device T is connected to the inspection server 20, the plate making server SV, and the offset printing press S.

  The system controller 11 is a PLC (Programmable Logic Controller) that receives various signals from the printing machine S and various parts inside the inspection device T and controls each part of the inspection device T.

  The pipeline image processing unit 12 includes four sets of image processing boards (a front image processing board 121A, a front image processing board 122A, a back image processing board 121B, and a back image). The processing board 122B) is mounted. The operations of the front image processing board 121A and the front image processing board 122A are alternately switched (the same applies to the back image processing board 121B and the back image processing board 122B). Each image processing board has a plurality of parallel processing chips mounted thereon and performs, for example, processing of an image of 10 million × 3 colors every 45 msec. The front image processing board 121A and the front image processing board 122A receive a camera signal from the front camera 31A via the front camera branch board 14, and obtain a high-resolution image of the surface based on the pulse signal received from the encoder branch board 16. Is generated (refer to Japanese Patent No. 311565 for a specific generation method), and a 100% inspection (inspection for all printed materials) is performed using the high-resolution image. Similarly, the back image processing board 121B and the back image processing board 122B also receive a camera signal from the back camera 31B via the back camera branch board 15, and perform a 100% inspection.

  Next, the contents of the 100% inspection will be described. The pipeline image processing unit 12 performs a 100% inspection based on an image (may be referred to as an image to be inspected) captured after the start of the 100% inspection, using a high-resolution image determined in advance by the operator as having no problem as a reference image. Specifically, a defect is detected by comparing a difference image obtained by subtracting the reference image from the inspection image or a difference image obtained by subtracting the inspection image from the reference image with the threshold image.

  When performing the inspection for comparing the image to be inspected with the reference image, the pipeline image processing unit 12 performs alignment of the two images so that the comparison can be appropriately performed. For this purpose, the pipeline image processing unit 12 divides an image into a plurality of blocks, and performs a shift amount calculation process of calculating a shift amount between both images for each block. In the shift amount calculation process, the pipeline image processing unit 12 divides the reference image and the image to be inspected into blocks, and uses a template image created for each of the blocks based on the reference image to divide the corresponding block of the image to be inspected. A matching part is searched for, and a difference between the matched coordinate and the coordinate of the template image is calculated as a deviation amount.

  In the first embodiment, in which the printing object is a continuous sheet, the pipeline image processing unit 12 creates one template image for one block, and searches the corresponding block of the inspection image with one template image. I do. Note that the template image creation processing will be described later using a flowchart.

  In the template image creation process, the pipeline image processing unit 12 may not be able to create a template image for a block having a small number of characteristic portions. The pipeline image processing unit 12 calculates the shift amount based on the template image for the block in which the template image was able to be created (“block with template”), and on the other hand, the block in which the template image was not created (“block without template”). )), The shift amount is determined on the basis of the shift amount of the upper, lower, left and right blocks or the block in the same row / column in which the shift amount is determined.

  When comparing the template image and the image to be inspected for each block, the pipeline image processing unit 12 creates a block image for the image to be inspected for each block in the same manner as the template image, and generates the block image using the template image. , The corresponding block image is searched for a matching portion. By comparing the images created by the same method as described above, a matching point can be more accurately searched.

  The pipeline image processing unit 12 performs the position correction of the image to be inspected based on the shift amount calculated for each block, thereby aligning the reference image and the image to be inspected.

  The parallel image processing unit 13 is equipped with a total of two image processing boards (a front image processing board 131A and a back image processing board 131B), one for the front side and one for the back side. 12 and operates in parallel. Each image processing board has a plurality of parallel processing chips mounted thereon and performs, for example, processing of an image of 10 million × 3 colors every 45 msec. The front image processing board 131A receives a camera signal from the front camera 31A via the front camera branch board 14, generates a high-resolution image of the front surface based on the pulse signal received from the encoder branch board 16, and generates plate making data ( A collation inspection with the image represented by the plate making data) and a double body inspection are performed. Plate making data is the original data used when creating a printing plate to be attached to a plate cylinder. In comparison inspection with plate making data, imposition, dimensions, proofreading, plate scratches, dust (with dust on the plate cylinder or blanket cylinder) Check if there is any).

  On the other hand, the double-body inspection is an inspection for comparing high-resolution images (that is, continuous high-resolution images) of two sheets (prints) printed when the blanket cylinder makes one rotation. For example, when the images do not match, a defect in which a part of the blanket cylinder has scratches or dust is detected.

  The front camera branch board 14 branches and sends the camera signal from the front camera 31A to the front image processing board 121A, the front image processing board 122A, and the front image processing board 131A at the same time.

  The back camera branch board 15 branches and sends the camera signal from the back camera 31B to the back image processing board 121B, the back image processing board 122B, and the front image processing board 131B at the same time.

  The encoder branch board 16 is attached to the plate cylinder of the offset printing press S, and is an encoder pulse signal (A, B, Z phase) output from an encoder 33 that is interlocked with the plate cylinder (interlocked with the rotation of the plate at 1: 1). And the encoder pulse signals (A, B, Z phases) of 0.05 to 0.1 mm are branched and output to the pipeline image processing unit 12 and the parallel image processing unit 13.

  The HUB 17 is a HUB for LAN connection with the system controller 11, the pipeline image processing unit 12, the parallel image processing unit 13, the inspection server 20, the plate making server SV, and the printing machine S.

  The inspection server 20 has a function of storing a reference image, storing a defective image when detecting a defect, and storing operation records and inspection log data when the inspection is started or stopped. The inspection server 20 is used when viewing inspection results via the LAN as needed. The plate-making server SV stores plate-making data and the like. When the operator sets the printing machine S to print a new printed material, the contents of the setting are acquired and the inspection device T performs inspection. Necessary information (including plate making data) is set in the inspection device T.

[1.2. Operation of template image creation processing for continuous sheet]
The continuous sheet template image creation processing by the pipeline image processing unit 12 will be described with reference to FIG. FIG. 3 is a flowchart illustrating an example of a continuous sheet template image creation process.

  First, the pipeline image processing unit 12 acquires a reference image obtained by capturing an image of a printed matter determined by the operator as a product without any problem (step S1). The reference image is an RGB image having RGB (Red, Green, Blue) components for each pixel, and each component is a value indicating a gradation indicating a density level (for example, “0 to 255 if 256 gradations”). ").

  Next, for the reference image, the pipeline image processing unit 12 acquires the luminance of each pixel corresponding to the blank section (blank section luminance) for each of the RGB (step S2). The blank paper portion is a paper portion to which no ink is attached.

  Next, the pipeline image processing unit 12 calculates a correction coefficient for each of the RGB so that the blank-sheet portion luminance becomes the target blank-sheet portion luminance for the reference image, and corrects each pixel (step S3). It is assumed that the target blank portion luminance is set in advance by an operator or the like, for example.

  Next, the pipeline image processing unit 12 divides the reference image (for example, an image obtained by capturing a printed matter having a size of 1030 mm in the width direction × 625 mm in the conveyance direction) into a plurality of blocks (Step S4). The number of blocks to be divided is arbitrarily set by the operator. For example, as shown in FIG. 4, the reference image 200 is divided into 512 blocks of 64 rows and 8 columns. However, it is preferable to change the number of lines according to the size of one pitch of the paper (80 pixels per line). The pipeline image processing unit 12 stores the range of each block (XY coordinates of the four corners) in a storage unit (not shown) in the pipeline image processing unit 12.

  Next, the pipeline image processing unit 12 performs edge extraction on the reference image for each of the RGB using a Sobel filter or an edge extraction filter (step S5). Specifically, the reference image (RGB) is divided into an R component image, a G component image, and a B component image, and edge extraction is performed for each image. The reason why edge extraction is performed for each RGB is to extract more edges. The image from which the edges are extracted (edge image) is hereinafter referred to as “A image”. 5 shows an example of an A image of an R component image, FIG. 6 shows an example of an A image of a G component image, and FIG. 7 shows an example of an A image of a B component image. Note that an edge is a point in the image where the density is rapidly changing, and corresponds to, for example, an outline or a line of an object in the image.

  Next, the pipeline image processing unit 12 binarizes the A image with a gradation a (“90” in the present embodiment) (Step S6). Specifically, each of the A image of the R component image, the A image of the G component image, and the A image of the B component image generated in the process of step S5 is binarized by the gradation a. The image obtained by the binarization is hereinafter referred to as “B image”. 8A shows an example of a B image of an R component image, FIG. 8B shows an example of a B image of a G component image, and FIG. 8C shows an example of a B image of a B component image. FIG. 8D shows an example of an RGB component B image obtained by combining these B images.

  Next, the pipeline image processing unit 12 binarizes the A image with the gradation b (“130” in the present embodiment) (Step S7). Specifically, each of the A image of the R component image, the A image of the G component image, and the A image of the B component image generated in the process of step S5 is binarized by the gradation b. The image obtained by the binarization is hereinafter referred to as “C image”. 9A illustrates an example of a C image of an R component image, FIG. 9B illustrates an example of a C image of a G component image, and FIG. 9C illustrates an example of a C image of a B component image. For example, FIG. 9D is a C component image of an RGB component obtained by combining these C images.

  Next, the pipeline image processing unit 12 horizontally differentiates the C image (extracts only the horizontal line of the edge of the symbol by horizontal differentiation) and then expands the image by two pixels in the horizontal direction (hereinafter referred to as “C1 image”). ") (Step S8). The expansion in the horizontal direction by two pixels means that the entire image is expanded in the horizontal direction. The expansion in the horizontal direction makes it easier to find the intersection in the processing of step S10 described below. FIG. 10A is an example of the C1 image.

  Next, after vertical differentiation of the C image, the pipeline image processing unit 12 generates an image expanded in the vertical direction by two pixels (hereinafter, referred to as a “C2 image”) (step S9). The expansion in the vertical direction by two pixels means that the entire image is expanded in the vertical direction. The expansion in the vertical direction makes it easier to find the intersection in the processing of step S10 described later. FIG. 10B is an example of a C2 image.

  Next, the pipeline image processing unit 12 combines the C1 image and the C2 image to generate a C intermediate image, and leaves only a straight line that is continuous over 8 pixels in both the horizontal and vertical directions and that intersects vertically. An image (hereinafter, referred to as “C3 image”) is generated, and the XY address of the intersection is obtained (step S10) Note that FIG.10 (C) is an example of an image obtained by combining the C1 image and the C2 image, and FIG. 12A is an example of a C3 image, and when there are a plurality of intersections, the respective XY addresses are acquired as shown in FIG.

  Next, the pipeline image processing unit 12 generates an image (hereinafter, referred to as “B1 image”) expanded horizontally by two pixels after the B image is horizontally differentiated (step S11). FIG. 11A is an example of a B1 image.

  Next, after the B image is vertically differentiated, the pipeline image processing unit 12 generates an image (hereinafter, referred to as a “B2 image”) expanded in the vertical direction by two pixels (step S12). FIG. 11B is an example of a B2 image.

  Next, the pipeline image processing unit 12 generates a B intermediate image by combining the B1 image and the B2 image, and leaves only a straight line that is continuous for at least 8 pixels in both the horizontal and vertical directions and that intersects vertically. An image (hereinafter, referred to as a “B3 image”) is generated (step S13) Note that FIG.11 (C) is an example of an image obtained by combining the B1 image and the B2 image, and FIG.12 (B) is an example of the B3 image. It is.

  Next, the pipeline image processing unit 12 generates a cutout image (C ′ image) of 32 pixels × 32 pixels centered on the XY address acquired in step S10 from the C3 image (see FIG. 12A). (Step S14). The size of the cut-out image (C ′ image) is arbitrary, and for example, a cut-out image (C ′ image) of 64 × 64 pixels may be generated. FIG. 12C is an example of a C ′ image.

  Next, the pipeline image processing unit 12 calculates a 32-pixel centered on the same XY address as the XY address centered when the cutout image is cut out from the B3 image (see FIG. 12B) in the process of step S13. A clipped image (B ′ image) of × 32 pixels is generated (step S15). Note that a cutout image (B ′ image) of 64 × 64 pixels may be generated (however, the same size as the C ′ image) as in the process of step S13. FIG. 12D is an example of a B ′ image.

  Next, if the difference between the C ′ image and the B ′ image is within the range of 1 pixel to 8 pixels, the pipeline image processing unit 12 determines the center addresses of these images (the XY addresses acquired in the processing in step S10). Is determined to be "template suitable" (step S16). When the difference between the C ′ image and the B ′ image is 0 pixel (completely coincident), it may be determined that “there is template aptitude”. On the other hand, if the difference between the C ′ image and the B ′ image is not within the range of 1 pixel to 8 pixels, the pipeline image processing unit 12 determines that the center addresses of these images are “no template suitability”. That is, if the difference between the C ′ image and the B ′ image is, for example, 9 pixels or more as in the present embodiment, there is too much difference, so that “No template aptitude” is determined. In the present embodiment, it is determined that “there is a template suitability” only when the difference is within the range of 1 pixel to 8 pixels. However, this is an example, and the threshold value may be appropriately changed.

  When a plurality of XY addresses are obtained in the process of step S10, the processes of steps S14 and S15 are performed by the number of the XY addresses to generate the C 'image and the B' image. At this time, the C 'image and the B' image cut out from the same XY address are stored as a set. Then, the process of step S16 is performed for each set of the C 'image and the B' image.

  Next, the pipeline image processing unit 12 selects one block divided in the process of step S4 (step S17).

  Next, the pipeline image processing unit 12 searches the block selected in the process of step S17 for an XY address determined to be "template suitable" (step S18). Specifically, a search is made for an XY address determined to be "template suitable" while shifting one pixel at a time in the X direction from a starting point (for example, a lower left corner) in the block. If it is not found even at the end of the block, it is shifted by one pixel in the Y direction, and again shifted by one pixel in the X direction, and a search is made for an XY address determined to be "template suitable". The search is terminated when one XY address determined as “template suitable” is found.

  Then, the pipeline image processing unit 12 determines whether or not there is an XY address “with template aptitude” in the block (Step S19). Specifically, if at least one XY address of “with template aptitude” is included in the block, “YES” is determined.

  If the pipeline image processing unit 12 determines that there is no XY address of “with template aptitude” in the block (step S19: NO), the pipeline image processing unit 12 determines that the block is a “block without template” (step S20). The process moves to the process in step S26. On the other hand, when the pipeline image processing unit 12 determines that there is an XY address of “with template aptitude” in the block (step S19: YES), it determines that the block is a “block with template” (step S21). ), The process proceeds to step S22.

  Next, the pipeline image processing unit 12 converts the A image into gradations ((a + b) / 2) for each of RGB in a manner similar to the processing in step S6 (“(90 + 130) / 2 = 110” in the present embodiment). Then, the same processing as that performed on the B image in the processing of steps S11 to S13 is performed to generate a D image (step S22). FIG. 13 is an example of a D image.

  Next, the pipeline image processing unit 12 cuts out 32 pixels × 32 pixels having the XY address “Template appropriate” as a center address from the D image, and generates a template image (D ′ image) (D ′ image in FIG. 13). 300 (see step S23). Note that the size of the D 'image may be 64 pixels × 64 pixels.

  Next, the pipeline image processing unit 12 checks the template image (D 'image) (Step S24). In this check, the D 'image is shifted by one pixel within the template verification range (a range of ± 64 pixels in the X direction and ± 128 pixels in the Y direction with reference to the center address of the template image (D' image)). It is verified that the difference between the D image and the D ′ image is one pixel or more at any position (excluding the center address). That is, it is verified that the template image (D 'image) is unique around it (within the template verification range).

  If the template image (D ′ image) is not unique within the template verification range as a result of the check, the pipeline image processing unit 12 determines that the template image (D ′ image) has failed, and returns to step S18. Then, another XY address of “with template aptitude” in the same block is searched, and the processing of steps S19 to S24 is performed for the XY address. If the result of the check does not change even if the processing of steps S18 to S24 is repeated twice, the processing shifts to the processing of step S20. On the other hand, when the template image (D 'image) is unique within the template verification range as a result of the check, the pipeline image processing unit 12 proceeds to the process of step S25.

  Next, the pipeline image processing unit 12 determines the identification information (for example, block ID) of one block selected in the processing of the latest step S17, the template image (D ′ image), and the center address (template image (D The 'image' is registered in a template table (not shown) in association with an XY address when the 'image' is cut out from the D image (step S25), and the process proceeds to step S26.

  After completing the processing in step S20 or step S25, the pipeline image processing unit 12 determines whether all blocks have been selected (step S26). That is, the pipeline image processing unit 12 determines whether or not all the blocks divided in the processing in step S4 have been selected in the processing in step S17. At this time, if it is determined that all the blocks have not been selected (step S26: NO), the pipeline image processing unit 12 shifts to the process of step S17, and selects one block that has not been selected. Select On the other hand, when it is determined that all the blocks have been selected (step S26: YES), the pipeline image processing unit 12 ends the continuous sheet template image creation processing.

  In the present embodiment, the pipeline image processing unit 12 cuts out an image (B ′ image) cut out from each of a B image obtained by binarizing the A image at the gradation a and a C image obtained by binarizing the A image at the gradation b. And the C ′ image) are compared, and when the difference is within a certain range (1 pixel to 8 pixels), the XY address at the time of cutting out the B ′ image and the C ′ image is determined to be appropriate for the template. . As a result, it is possible to find feature points in a certain gradation range (range of gradations a and b). Further, the pipeline image processing unit 12 generates a template image (D ′ image) from the D image obtained by binarizing the A image with the gradation (a + b) / 2, based on the XY address determined to be appropriate for the template. I do. Thereby, even if the gradation of the inspected image changes in the range of gradations a and b, it can be appropriately matched with the template image (D 'image).

[1.3. Operation of continuous sheet inspection processing]
Next, the continuous sheet inspection processing by the pipeline image processing unit 12 will be described with reference to FIG. FIG. 14 is a flowchart illustrating an example of the continuous sheet inspection process. The continuous sheet inspection process is a 100% inspection that performs a comparison process between the reference image and the image to be inspected.

  First, the pipeline image processing unit 12 performs a continuous sheet shift amount calculation process (step S31). The continuous sheet shift amount calculation processing will be described later.

  Next, the pipeline image processing unit 12 performs the position correction of the image to be inspected based on the shift amount calculated for each block by the continuous sheet shift amount calculation processing (step S32).

  Next, the pipeline image processing unit 12 performs an inspection for comparing the position-corrected image to be inspected with the reference image (step S33).

  Next, the pipeline image processing unit 12 outputs the result of the comparison inspection performed in the processing of Step S33 (Step S34), and ends the continuous sheet inspection processing. In the process of step S34, the pipeline image processing unit 12 creates, for example, result data indicating the result of the comparison inspection and transmits the result data to the inspection server 20 (a defective image at the time of defect detection, The test result may be transmitted together with the operation record or the test log data), or the test result may be displayed on a display device (for example, a display device connected to the test server 20).

[1.4. Operation of continuous sheet shift amount calculation processing]
Next, the continuous sheet shift amount calculation processing by the pipeline image processing unit 12 will be described with reference to FIG.

  First, the pipeline image processing unit 12 obtains an image to be inspected obtained by capturing an image of a printed material to be subjected to the 100% inspection (step S51). The image to be inspected is an RGB image, and each component is represented by a value indicating a gradation of “0 to 255”.

  Next, the pipeline image processing unit 12 acquires, for each of the RGB images, the luminance of each pixel corresponding to the blank section (blank section luminance) for the image to be inspected (step S52).

  Next, the pipeline image processing unit 12 calculates a correction coefficient for each of the R, G, and B components of the image to be inspected so that the blank-sheet luminance becomes the target blank-sheet luminance, and corrects each pixel (step S53). It is assumed that the same value as that in step S3 of the continuous sheet template image creation process is set in advance by the operator or the like as the target blank sheet portion brightness.

  Next, the pipeline image processing unit 12 divides the image to be inspected into a plurality of blocks as in the case of the reference image (step S54).

  Next, the pipeline image processing unit 12 performs edge extraction on the image to be inspected for each of the RGB using a Sobel filter, similarly to the reference image (step S55). Note that the image extracted in the continuous sheet inspection process is also referred to as “A image”.

  Next, the pipeline image processing unit 12 converts the A image into the gradation ((a + b) / 2) (in the present embodiment, “(90 + 130) / 2”, as in step S22 of the continuous sheet template image creation processing. = 110 ”), the same processing as that performed on the B image in the processing of steps S11 to S13 of the continuous sheet template image generation processing is performed on the image binarized to generate the D image (step S110). S56). Note that a and b use the same values as those in the continuous sheet template image creation processing.

  Next, the pipeline image processing unit 12 performs a continuous sheet / template block processing (step S57). The continuous sheet / template block processing will be described later. The continuous sheet / template block processing is processing for determining a shift amount for a template block.

  Next, the pipeline image processing unit 12 determines whether or not there is a block whose deviation amount has not been determined (step S58). In the continuous sheet / template block processing, the shift amount is determined for the template block, and the block for which the shift amount has not been determined corresponds to the template-less block. When determining that there is no block in which the shift amount has not been determined (step S58: NO), the pipeline image processing unit 12 ends the continuous sheet shift amount calculation processing. On the other hand, when the pipeline image processing unit 12 determines that there is a block whose deviation amount has not been determined (step S58: YES), the pipeline image processing unit 12 selects one block without template (step S59), and determines the deviation amount. The shift amount is determined based on the shift amount of the upper, lower, left and right blocks or the block in the same row / same column (step S60), and the process proceeds to step S58. Then, the pipeline image processing unit 12 repeats the processing of steps S58 to S60 until all “template-less blocks” are selected (until the deviation amounts are determined for all template-less blocks).

[1.5. Operation of block processing with continuous sheet template]
Next, block processing with a continuous sheet template by the pipeline image processing unit 12 will be described with reference to FIG.

  First, the pipeline image processing unit 12 selects one block determined as a “block with template” in the process of step S21 of the template image creation process for a continuous sheet (step S101).

  Next, the pipeline image processing unit 12 obtains a template image (D ′ image) and a center address included in the block selected in step S101 from the template table (see step S25 of the continuous sheet template image creation process). Is acquired (step S102).

  Next, the pipeline image processing unit 12 sets a predetermined range (for example, a range of ± 64 pixels in the X direction and ± 128 pixels in the Y direction) based on the center address acquired in the processing of step S102 as a reference. The D ′ image is shifted by one pixel with respect to the D image generated in the process of step S56, and the difference between the D image and the D ′ image is calculated (step S103).

  Next, the pipeline image processing unit 12 specifies the position where the difference is minimum (Step S104). When there are a plurality of positions having the same value with the smallest difference, one position closest to the center address is specified.

  Next, the pipeline image processing unit 12 determines the difference between the position specified in the processing in step S104 and the center address acquired in the processing in step S102 as the deviation amount (dx, dy) (step S105). Note that the determined shift amount is stored in association with identification information (for example, a block ID) of a block including the center address.

  Next, the pipeline image processing unit 12 determines whether or not all “blocks with template” have been selected (Step S106). At this time, if it is determined that all “blocks with template” have not been selected (step S106: NO), the pipeline image processing unit 12 shifts to the processing of step S101 and has selected it up to that point. One of the "blocks with template" is selected. Then, the processing of steps S101 to S106 is repeated until all “blocks with template” are selected. On the other hand, when it is determined that all “blocks with template” have been selected (step S106: YES), the pipeline image processing unit 12 ends the continuous sheet / template block processing and calculates the continuous sheet deviation amount. Return to processing.

[2. Second Embodiment]
Next, a second embodiment of the present invention will be described with reference to the drawings. Note that the second embodiment is an embodiment in which the present invention is applied to an inspection device installed in an offset sheet-fed printing press. Note that the second embodiment has many parts in common with the first embodiment, and therefore, the following description will focus on the differences.

[2.1. Configuration of offset sheet-fed printing press and inspection device]
The configuration of the offset sheet-fed printing press according to the second embodiment is not directly related to the present invention, and thus the description is omitted.

  Next, a configuration of the inspection device T according to the second embodiment will be described. The inspection device T according to the second embodiment has basically the same configuration as the inspection device T according to the first embodiment.

  The pipeline image processing unit 12 includes four sets of image processing boards (a front image processing board 121A, a front image processing board 122A, a back image processing board 121B, and a back image). The processing board 122B) is mounted. The operations of the front image processing board 121A and the front image processing board 122A are alternately switched (the same applies to the back image processing board 121B and the back image processing board 122B). Each image processing board has a plurality of parallel processing chips mounted thereon and performs, for example, processing of an image of 10 million × 3 colors every 180 msec. The front image processing board 121A and the front image processing board 122A receive a camera signal from the front camera 31A via the front camera branch board 14, and obtain a high-resolution image of the surface based on the pulse signal received from the encoder branch board 16. Is generated (refer to Japanese Patent No. 311565 for a specific generation method), and a 100% inspection (inspection for all printed materials) is performed using the high-resolution image. Similarly, the back image processing board 121B and the back image processing board 122B also receive a camera signal from the back camera 31B via the back camera branch board 15, and perform a 100% inspection.

  As in the first embodiment, the pipeline image processing unit 12 aligns the two images so that the comparison can be appropriately performed when performing the inspection for comparing the image to be inspected with the reference image in the 100% inspection. For this purpose, the pipeline image processing unit 12 divides an image into a plurality of blocks, and performs a shift amount calculation process of calculating a shift amount between both images for each block.

  In the second embodiment in which the substrate is a sheet, which is an example of a cut sheet, when each of the printed sheets is sequentially photographed by the camera 31, the sheets are placed in an irregular direction (vertical direction). (Horizontal or rotational direction), the direction of the continuous image to be inspected becomes irregular. Two template images are created for one block, and a search is made within the corresponding block of the image to be inspected with the two template images. Note that the template image creation processing will be described later using a flowchart. On the other hand, in the first embodiment in which the orientation (deviation) of successive images to be inspected is related, and the printing speed is high and the speed is also required for 100% inspection, as described above, one template is used for one block. Alignment is performed using images.

  In the template image creation processing, the pipeline image processing unit 12 may not be able to create two template images for a block having a small number of characteristic portions. The pipeline image processing unit 12 calculates a shift amount based on the two template images for a block in which two template images have been created (“block with template”) (calculates a shift amount for two coincident points). On the other hand, with respect to the blocks for which the template image could not be created (“blocks without template”), the upper, lower, left, and right blocks or the blocks in the same row / same column for which the amount of displacement is determined The shift amount is determined based on the shift amount.

  When comparing the template image and the image to be inspected for each block, the pipeline image processing unit 12 creates a block image for the image to be inspected for each block in the same manner as the template image, and generates the block image using the template image. , The corresponding block image is searched for a matching portion. By comparing the images created by the same method as described above, a matching point can be more accurately searched.

  The pipeline image processing unit 12 performs the position correction of the image to be inspected based on the shift amount calculated for each block, thereby aligning the reference image and the image to be inspected.

[2.2. Operation of Sheet Image Template Image Creation Processing]
With reference to FIG. 17, a description will be given of the sheet template image creation processing by the pipeline image processing unit 12. FIG. 17 is a flowchart illustrating an example of a sheet-sheet template image creation process.

  The process at the beginning of the sheet template image creation process is the same as Steps S1 to S16 of the continuous sheet template image creation process (see FIG. 3).

  After completing the processing in step S16, the pipeline image processing unit 12 selects one block divided in the processing in step S4 (step S201).

  Next, the pipeline image processing unit 12 searches the block selected in the process of step S17 for an XY address determined to be "template suitable" (step S202). Specifically, a search is made for an XY address determined to be "template suitable" while shifting one pixel at a time in the X direction from a starting point (for example, a lower left corner) in the block. If it is not found even at the end of the block, it is shifted by one pixel in the Y direction, and again shifted by one pixel in the X direction, and a search is made for an XY address determined to be "template suitable". The search is terminated when two XY addresses determined as “template suitable” are found.

  Then, the pipeline image processing unit 12 determines whether or not there are two XY addresses “with template aptitude” in the block (Step S203). If the pipeline image processing unit 12 determines that there are no two XY addresses of “with template aptitude” (step S203: NO), it determines that the block is a “template-less block” (step S204). The process proceeds to S210. On the other hand, when the pipeline image processing unit 12 determines that there are two XY addresses of “with template aptitude” (step S203: YES), the block is determined to be a “block with template” (step S205). Then, the process proceeds to the process of step S206.

  Next, the pipeline image processing unit 12 converts the A image into the gradation ((a + b) / 2) (in the present embodiment, “(90 + 130) / 2”, as in step S22 of the continuous sheet template image creation processing. = 110 ”), the same processing as that performed on the B image in the processing of steps S11 to S13 of the continuous sheet template image generation processing is performed on the image binarized to generate the D image (step S110). S206).

  Next, the pipeline image processing unit 12 cuts out 32 pixels × 32 pixels centered on the XY address “Template appropriate” from the D image, and generates a template image (D ′ image) (Step S207). In the second embodiment, since there are two XY addresses of “with template aptitude” for one block, the pipeline image processing unit 12 performs the process of step S207 for each block, and executes two template images (D ′). Image).

  Next, the pipeline image processing unit 12 checks the template image (D 'image) (Step S208). In this check, the D 'image is shifted by one pixel within the template verification range (a range of ± 64 pixels in the X direction and ± 128 pixels in the Y direction with reference to the center address of the template image (D' image)). At any position (excluding the center address), it is verified that the difference between the D image and the D ′ image is one or more pixels. That is, it is verified that the template image (D 'image) is unique around it (within the template verification range). In the second embodiment, since there are two D 'images for one block, the pipeline image processing unit 12 verifies each D' image in the process of step S208.

  If the template image (D ′ image) is not unique within the template verification range as a result of the check, the pipeline image processing unit 12 determines that the template image (D ′ image) has failed, and The process returns to step S202 in order to replace the XY address, and another XY address of “with template aptitude” in the same block is searched, and the processes of steps S203 to S208 are performed on the XY address. If the two template images (D 'images) do not pass even after performing the processes of steps S202 to S208 twice, the process proceeds to step S204. On the other hand, as a result of the check, the pipeline image processing unit 12 passes if the template image (D ′ image) is unique within the template verification range. When the two template images (D 'images) pass, the pipeline image processing unit 12 proceeds to the process of step S209.

  Next, the pipeline image processing unit 12 determines the identification information (for example, block ID) of one block selected in the process of the latest step S201, two template images (D ′ images), and each template image (D ′ image). The center address of the D 'image), its average coordinates, and the center address of the block are registered in a template table (not shown) in association with each other (step S209), and the process proceeds to step S210.

  After completing the processing in step S204 or step S209, the pipeline image processing unit 12 determines whether all blocks have been selected (step S210). That is, the pipeline image processing unit 12 determines whether or not all the blocks divided in the processing in step S4 have been selected in the processing in step S201. At this time, when it is determined that all the blocks have not been selected (step S210: NO), the pipeline image processing unit 12 proceeds to the process of step S201, and selects one block that has not been selected. Select Then, the processing of steps S201 to S210 is repeated until all blocks are selected. On the other hand, when determining that all the blocks have been selected (step S210: YES), the pipeline image processing unit 12 ends the sheet-sheet template image creation processing.

  In the second embodiment, the pipeline image processing unit 12 cuts out an image (B ′) cut out from each of a B image obtained by binarizing the A image with the gradation a and a C image obtained by binarizing the A image with the gradation b. Image and C ′ image), and when the difference is within a certain range (1 pixel to 8 pixels), the XY address at the time of cutting out the B ′ image and C ′ image is determined to be appropriate for the template. I do. As a result, it is possible to find feature points in a certain gradation range (range of gradations a and b). In addition, the pipeline image processing unit 12 converts the template image (D ′ image) into two based on the XY address determined as having the template suitability from the D image obtained by binarizing the A image with the gradation (a + b) / 2. Generate one. Thereby, even if the gradation of the image to be inspected changes in the range of gradations a and b, it can be appropriately matched with the template image (D ′ image), and two template images per block can be used. Since the displacement is calculated and the position is corrected, the position can be corrected with higher accuracy than in the first embodiment.

[2.3. Operation of Sheet Inspection Processing]
Next, the sheet inspection processing by the pipeline image processing unit 12 will be described with reference to FIG. FIG. 18 is a flowchart illustrating an example of the sheet inspection process. The sheet inspection process is a 100% inspection that performs a comparison process between the reference image and the image to be inspected.

  First, the pipeline image processing unit 12 performs a sheet / shift amount calculation process (step S231). The sheet / shift amount calculation processing will be described later.

  Next, the pipeline image processing unit 12 performs the same processing as Steps S32 to S34 of the continuous sheet inspection processing (see FIG. 14), and ends the sheet inspection processing.

[2.4. Operation of Sheet / Shift Amount Calculation Processing]
Next, the sheet / displacement amount calculation processing by the pipeline image processing unit 12 will be described with reference to FIG.

  The process at the beginning of the sheet / shift amount calculation process is the same as Steps S51 to S56 of the continuous sheet shift amount calculation process (see FIG. 15).

  Next, the pipeline image processing unit 12 performs a sheet / template presence block process (step S251). The sheet / template block processing will be described later.

  Next, the pipeline image processing unit 12 performs the same processing as Steps S58 to S60 in the continuous sheet shift amount calculation processing (see FIG. 15), and ends the sheet / shift amount calculation processing.

[2.5. Operation of block processing with sheet / template]
Next, the sheet / template block processing by the pipeline image processing unit 12 will be described with reference to FIG.

  First, the pipeline image processing unit 12 selects one block determined as a “block with template” in the processing of step S205 of the sheet-sheet template image creation processing (step S301).

  Next, the pipeline image processing unit 12 acquires two sets of the template image (D 'image) and the center address included in the block selected in the processing of step S301 from the template table (step S302).

  Next, the pipeline image processing unit 12 sets a predetermined range (a range of ± 64 pixels in the X direction and ± 128 pixels in the Y direction) based on the first set of center addresses acquired in the process of step S302. Is shifted from the D image by one pixel in the first set of D 'images, and the difference between the D image and the D' image is calculated (step S303).

  Next, the pipeline image processing unit 12 specifies the position where the difference is minimum (Step S304). When there are a plurality of positions having the same value with the smallest difference, one position closest to the center address is specified.

  Next, the pipeline image processing unit 12 sets a predetermined range (a range of ± 64 pixels in the X direction and a range of ± 128 pixels in the Y direction) based on the center address of the second set acquired in the process of step S302. Is shifted by one pixel from the D image with respect to the D image, and the difference between the D image and the D 'image is calculated (step S305).

  Next, the pipeline image processing unit 12 specifies the position where the difference is minimum (Step S306). When there are a plurality of positions having the same value with the smallest difference, one position closest to the center address is specified.

  Next, the pipeline image processing unit 12 calculates a shift amount for each of the two template images (D 'images) (step S307). Specifically, the difference between the position specified in the process of step S304 and the center address of the first set acquired in the process of step S302 is calculated as the first shift amount (dx1, dy1), and the process of step S306 is performed. Then, the difference between the position specified in and the second set of center addresses acquired in the processing of step S302 is calculated as the second shift amount (dx2, dy2).

  Next, the pipeline image processing unit 12 determines the average value of the two shift amounts calculated in the process of step S307 as the shift amount (dx, dy) of the block (step S308). Note that the determined shift amount is stored in association with identification information (for example, block ID) of the block.

  Next, the pipeline image processing unit 12 determines whether or not all “blocks with template” have been selected (Step S309). At this time, if it is determined that all the “blocks with template” have not been selected (step S309: NO), the pipeline image processing unit 12 proceeds to the process of step S301, and has selected it. One of the "blocks with template" is selected. Then, the processing of steps S301 to S309 is repeated until all "blocks with template" are selected. On the other hand, if it is determined that all “blocks with template” have been selected (step S309: YES), the pipeline image processing unit 12 ends the sheet / template block processing and returns to sheet / shift. The process returns to the amount calculation process.

  As described above, the inspection device T (an example of a “feature image generation device”) of the first embodiment and the second embodiment includes the pipeline image processing unit 12 (“edge extraction unit”, “binarization unit”). , The “first combining unit”, the “intersection acquisition unit”, the “second combining unit”, the “cut-out image generating unit”, and the “feature image generating unit” are examples), and the A image (““ C image (an example of a “first binarized image”) in which an A image is generated, and the A image is binarized with a gray level b (an example of a “first gray level”); Image (an example of a “second binarized image”) obtained by binarizing an image with a gradation a (an example of a “second gradation”) lower than the gradation b, and an edge image A D image (an example of a “third binarized image”) binarized at an intermediate gradation of the tone b, and a C1 image (“first water”) obtained by extracting a horizontal straight line from the C image Image) and a C2 image (an example of a “first vertical image”) obtained by extracting a vertical straight line from the C image to generate a C3 image (an example of a “first synthesized image”). XY addresses (an example of “intersection coordinates”) of a horizontal straight line and a vertical straight line are obtained, and a B1 image (an example of a “second horizontal image”) obtained by extracting a horizontal straight line from the B image and a vertical line from the B image B2 image (an example of a “second vertical image”) obtained by extracting the straight line of “1”, a B3 image (an example of a “second synthesized image”) is generated, and an image including an XY address is cut out from the C3 image and the B3 image. To generate a C ′ image (an example of a “first clipped image”) and a B ′ image (an example of a “second clipped image”), respectively. When the pixel is within the range of the pixel (an example of “when the predetermined condition is satisfied”) To generate a template image cut out an image including an XY address from D images (D 'image. An example of the "feature image").

  Therefore, according to the inspection device T, the C ′ image obtained from the image obtained by binarizing the A image obtained by extracting the edge from the reference image with the gradation b and the image obtained by binarizing the edge image with the gradation a Is the same position from the D image obtained by binarizing the A image with a gray level between the gray level a and the gray level b when the difference pixel number of the B ′ image obtained from is within the range of 1 pixel to 8 pixels. Is cut out as a template image (D ′ image), and if the gradation in the edge image of the comparison image to be compared with the reference image is between gradation b and gradation a, the positions of the reference image and the image to be inspected are Matching can be performed. That is, a template image (D 'image) used for matching for aligning the high-resolution reference image and the image to be inspected can be appropriately generated from the reference image.

  As described above, since the template image (D ′ image) is generated based on the edges detected in both the image (B image and C image) binarized by the gradation a and the gradation b, Even if there is a change such as a decrease in the overall gradation during the printing process, the edge on which the template image is generated is detected when the gradation changes within the range of gradations a and b. Since it becomes possible, position correction can be performed. In other words, it is guaranteed that proper alignment is performed even if there is a gradation change in the edge portion within the range of gradations a and b. If a template image is generated based on an edge detected in an image binarized by one gradation, it is not known in which gradation range the edge can be detected, and during the printing process, the edge is not detected. If the gray level changes to an undetectable gray level, alignment cannot be performed suddenly, and the inspection must be stopped.

  Note that the pipeline image processing unit 12 of the inspection apparatus T generates an A image in which edges are extracted from the reference image, and a C image obtained by binarizing the A image with the gradation b and an edge image with a gradation b A B image binarized with a low gradation a and a D image obtained by binarizing an edge image with a gradation between the gradation a and the gradation b are generated, and a horizontal straight line and a vertical An XY address (an example of “intersection coordinates”) of a straight line is obtained, an image including the XY address is cut out from the C image and the B image, and a C ′ image (an example of the “first cut image”) and a B ′ image ( An example of the “second clipped image” is generated, and when the number of difference pixels between the C ′ image and the B ′ image is within the range of 1 to 8 pixels (an example of “when the predetermined condition is satisfied”), A template image (D ′ image obtained by cutting out an image including XY addresses from the D image. It may be generated an example) of a feature image. "

  In the inspection apparatuses T of the first and second embodiments, the pipeline image processing unit 12 (an example of a “division unit”) divides a reference image into a plurality of blocks, and generates a template image (D ′ image) for each block. ). This makes it possible to calculate the amount of deviation for each block between the reference image and the image to be inspected, and thus to perform alignment for each block.

  Further, in the inspection device T of the first embodiment, when the pipeline image processing unit 12 captures a part of the continuous sheet while the reference image is transporting the continuous sheet along the longitudinal direction, Next, one template image (D ′ image) is generated for one block.

  Furthermore, in the inspection device T of the second embodiment, when the pipeline image processing unit 12 captures a sheet (an example of a “cut sheet”), Two template images (D 'images) are generated. When an image is taken after printing a sheet, the sheet is irregularly displaced in the vertical, horizontal, and rotational directions. However, according to the inspection apparatus T of the second embodiment, two template images per block are used. By calculating the amount of displacement, it is possible to perform positioning with high accuracy.

  Furthermore, in the inspection apparatus T of the first embodiment and the second embodiment, the pipeline image processing unit 12 (an example of the “determination unit”) uses the template verification range (“verification area”) based on the XY address in the D image. “In” is compared with the D image while moving the template image (D ′ image), and when the template image (D ′ image) matches at a position other than the XY address, the template image (D ′ image) 'Image). As a result, an image without a pattern (feature) similar to the surroundings (within the template verification range) can be adopted as the template image (D 'image).

  In addition to the image comparison processing performed for inspecting the printed matter, when performing the image comparison processing for various purposes, appropriate comparison is performed because one of the images is distorted, stretched / shrinked, rotated, or the like. If it is not possible, the deviation amount calculation processing of the present invention can be used to align the two images.

  In addition, the template image (D ′ image) based on the image (D image) binarized by the middle gradation ((a + b) / 2) of the gradation a and the gradation b in the first embodiment and the second embodiment. ) Is generated, but it is not always necessary to generate the template image based on the binarized image at the middle gradation. The image (D image) from which the template image is cut out may be an image binarized with a gray level shifted higher or lower than the middle gray level, or the gray level of the image to be inspected may decrease during the printing process. If so, the image may be a binarized image with a higher gradation than the middle.

T inspection apparatus 11 system controller 12 pipeline image processing unit 13 parallel image processing unit 14 front camera branch board 15 back camera branch board 16 encoder branch board 17 HUB
31A Front camera 31B Back camera 32A Front illumination 32B Back illumination 33 Encoder S Offset rotary printing press U1-U4 Printing unit U5 Dryer U6 Folding machine

Claims (10)

An edge extraction unit that generates an edge image obtained by extracting an edge from the reference image,
A first binarized image obtained by binarizing the edge image with a first gradation, and a second binarized image obtained by binarizing the edge image with a second gradation lower than the first gradation A binarizing unit configured to generate an image and a third binarized image obtained by binarizing the edge image with a third gradation between the first gradation and the second gradation;
An intersection acquisition unit that acquires intersection coordinates of a horizontal straight line and a vertical straight line in the first binarized image;
A cutout image generation unit that cuts out an image including the intersection coordinates from the first binarized image and the second binarized image, and generates a first cutout image and a second cutout image, respectively;
A feature image generating unit configured to generate a feature image obtained by cutting out an image including the intersection coordinates from the third binarized image when a difference pixel number between the first cutout image and the second cutout image satisfies a predetermined condition; ,
A characteristic image generating apparatus comprising: The feature image generation device according to claim 1,
A first horizontal image obtained by extracting a horizontal straight line from the first binary image and a first vertical image obtained by extracting a vertical straight line from the first binary image are combined to generate a first combined image. 1 synthesis unit,
A second horizontal image obtained by extracting a horizontal straight line from the second binary image and a second vertical image obtained by extracting a vertical straight line from the second binary image are combined to generate a second combined image. 2 synthesis unit,
Further comprising
The intersection obtaining unit,
Acquiring the intersection coordinates based on the first combined image generated from the first binarized image;
The cut-out image generation unit,
The first cutout image and the second cutout image are generated from the first combined image generated from the first binary image and the second combined image generated from the second binary image, respectively. A feature image generation device characterized by the above-mentioned. The characteristic image generation device according to claim 2,
The first synthesizing unit extracts a horizontal straight line from the first binary image, extracts the first horizontal image enlarged in the horizontal direction, and extracts a vertical straight line from the first binary image, Combining the first vertical image enlarged in the vertical direction to generate the first combined image;
The second synthesizing unit extracts a horizontal straight line from the second binary image, extracts the second horizontal image enlarged in the horizontal direction, and extracts a vertical straight line from the second binary image. A characteristic image generation apparatus, wherein the second vertical image enlarged in the vertical direction is synthesized to generate the second synthesized image. It is a characteristic image generation device according to any one of claims 1 to 3,
The image processing apparatus further includes a dividing unit that divides the reference image into a plurality of blocks ,
Prior Symbol feature image generating unit, said speed difference pixels to generate the feature image for the block including the intersection coordinates at the time of cutting out the predetermined condition is satisfied the first cut-out image and the second cut-out image A feature image generation device to be characterized. The feature image generation device according to claim 4, wherein
When the reference image is an image obtained by photographing a part of the continuous sheet while the continuous sheet is being conveyed along the longitudinal direction, one feature image is generated for one block. Characteristic image generation device. The feature image generation device according to claim 4, wherein
A feature image generating apparatus, wherein, when the reference image is an image obtained by photographing a cut sheet obtained by cutting a continuous sheet into individual sheets, two feature images are generated for one block. The characteristic image generation device according to claim 1, wherein:
In the verification area based on the intersection coordinates in the third binary image, the feature image is compared with the third binary image while moving the feature image. A characteristic image generating apparatus, further comprising: a determining unit that determines a characteristic image as an inappropriate characteristic image. An inspection device that performs position correction of an inspection image using the characteristic image created by the characteristic image generation device according to any one of claims 1 to 7, and performs a comparative inspection between the reference image and the inspection image. And
A deviation amount calculation unit that specifies coordinates corresponding to the feature image in the inspection image and calculates a deviation amount that is a difference from the intersection coordinates acquired by the intersection acquisition unit.
A position correction unit that performs position correction of the image to be inspected based on the displacement amount;
A comparing unit that compares the inspected image and the reference image subjected to the position correction,
An inspection apparatus comprising: An edge extraction step of generating an edge image obtained by extracting an edge from the reference image,
A first binarized image obtained by binarizing the edge image with a first gradation, and a second binarized image obtained by binarizing the edge image with a second gradation lower than the first gradation A binarization step of generating an image and a third binarized image obtained by binarizing the edge image with a third gradation between the first gradation and the second gradation;
An intersection acquisition step of acquiring intersection coordinates of a horizontal straight line and a vertical straight line in the first binarized image;
A cutout image generation step of cutting out an image including the intersection coordinates from the first binarized image and the second binarized image to generate a first cutout image and a second cutout image, respectively;
A feature image generating step of generating a feature image obtained by cutting out an image including the intersection coordinates from the third binarized image when a difference pixel number between the first cutout image and the second cutout image satisfies a predetermined condition; ,
A feature image generation method characterized by including: Computer
An edge extraction unit that generates an edge image obtained by extracting an edge from the reference image,
A first binarized image obtained by binarizing the edge image with a first gradation, and a second binarized image obtained by binarizing the edge image with a second gradation lower than the first gradation A binarization unit that generates an image and a third binarized image obtained by binarizing the edge image with a third gradation between the first gradation and the second gradation;
An intersection acquisition unit for acquiring intersection coordinates of a horizontal straight line and a vertical straight line in the first binarized image;
A cutout image generation unit that cuts out an image including the intersection coordinates from the first binarized image and the second binarized image, and generates a first cutout image and a second cutout image, respectively;
A feature image generation unit configured to generate a feature image obtained by cutting out an image including the intersection coordinates from the third binarized image when a difference pixel number between the first cutout image and the second cutout image satisfies a predetermined condition;
A feature image generation program characterized by functioning as: