Inspection device and its control method

Abstract

To solve such a problem that erroneous detection occurs due to noise included in a scan image when a reference image and a scan image are brought close to each other with color matching.SOLUTION: An inspection device for inspecting an image formed on a recording medium by a printer stores image data used for forming the image on the recording medium as a reference image, acquires the image data of an inspection object formed on the recording medium, performs alignment between the reference image applied with a noise component and the image data of the inspection object, and performs collation processing between the aligned reference image and image data of the inspection object.SELECTED DRAWING: Figure 5

Description

Technical Field

[0001] The present invention relates to an inspection apparatus and its control method. In the law

Background Art

[0002] In printed matter printed and output by a printing apparatus, coloring materials such as ink and toner may adhere to unintended locations, resulting in stains. Alternatively, insufficient coloring material may fail to adhere to the locations where an image should be formed, causing color bleeding where the color is lighter than normal. Such so-called printing abnormalities, including stains and color bleeding, degrade the quality of the printed matter. Therefore, it is necessary to inspect the printed matter for abnormalities to ensure its quality.

[0003] Visual inspection by an inspector to check for printing abnormalities requires a lot of time and cost. In recent years, inspection systems that perform inspections automatically without relying on visual inspection have been proposed. Specifically, it involves aligning a digital image (reference image) used for printing with scan image data obtained by scanning the printed matter (hereinafter referred to as a scan image), and performing image comparison and determination processing for the presence or absence of abnormalities to determine the image quality.

[0004] Patent Document 1 describes a method of performing an inspection by converting a reference image drawn in the CMYK color space into the same RGB color space as the scan image. According to the method described in this Patent Document 1, in the comparison between the reference image, which is a digital image, and the scan image, the occurrence of false detection is suppressed by taking into account errors caused by the state of the scanner and the accuracy of color conversion.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] However, simply bringing the reference image and the scanned image closer together through color matching can lead to false detections due to noise. This is because the reference image is a uniform image as it is digital data, while the scanned image is image data that contains a lot of noise, such as variations in the surface texture and transmittance of the paper, and the signal-to-noise ratio of the scanner. Therefore, if the reference image and the scanned image are compared directly, the difference will be amplified in some areas, leading to false detections.

[0007] The object of the present invention is to solve at least one of the problems of the prior art described above.

[0008] The object of the present invention is to provide a technology that suppresses false detections in the matching of a reference image with a scanned image of the object to be inspected, and improves inspection accuracy. [Means for solving the problem]

[0009] To achieve the above objective, an inspection apparatus according to one aspect of the present invention has the following configuration. That is, An inspection device for inspecting images formed on a recording medium by a printing device, A storage means for storing the image data used to form the image as a reference image on the recording medium, An acquisition means for acquiring image data of an object to be inspected formed on the aforementioned recording medium, A means for adding noise components to the aforementioned reference image, A positioning means for aligning the reference image to which the noise component has been added by the aforementioned adding means with the image data to be inspected, A matching means that performs a matching process between a reference image aligned by the alignment means and the image data of the object to be inspected, A means for determining the variance of pixel values ​​obtained by optically reading a recording medium in which no image has been formed, to have death, The imparting means modifies the noise component based on the variance. It is characterized by doing so. [Effects of the Invention]

[0010] According to the present invention, there is an effect that false detection can be suppressed and inspection accuracy can be improved in the collation between a reference image and a scan image of an inspection target.

[0011] Other features and advantages of the present invention will become apparent from the following description with reference to the accompanying drawings. In the accompanying drawings, the same or similar configurations are denoted by the same reference numerals.

Brief Description of the Drawings

[0012] The accompanying drawings are included in the specification, form a part thereof, show embodiments of the present invention, and are used to explain the principles of the present invention together with the description. [Figure 1] The figure which shows the structural example of the inspection system containing the inspection apparatus which concerns on Embodiment 1 of this invention. [Figure 2] The block diagram explaining the hardware constitution of the image forming apparatus which concerns on Embodiment 1. [Figure 3] The figure explaining the structure of the printer part of the image forming apparatus which concerns on Embodiment 1. [Figure 4] The schematic diagram (A) explaining the internal structure of the inspection apparatus which concerns on Embodiment 1, and the top view (B) of the conveyance belt seen from the inspection sensor side. [Figure 5] The block diagram explaining the functional structure of the inspection apparatus control part of the inspection apparatus which concerns on Embodiment 1. [Figure 6] The flowchart explaining the inspection processing by the inspection apparatus which concerns on Embodiment 1. [Figure 7] The figure which shows an example of the random number map which concerns on Embodiment 1. [Figure 8] The figure which shows an example which divided the reference image which concerns on Embodiment 1. [Figure 9] The figure explaining the four corner areas of the print output paper in Embodiment 3. [Figure 10] The figure which shows an example of the UI screen displayed on the operation part / display part of the inspection apparatus which concerns on Embodiment 1. [Figure 11] It is a flowchart explaining the collation processing by the inspection apparatus which concerns on Embodiment 1. [Figure 12] Schematic diagram of neighborhood search according to Embodiment 1. [Figure 13] Diagram (A) showing an example of a reference image according to an embodiment, diagram (B) showing an affine transformation formula, and diagram (C) showing an example of a CMYK_to_RGB lookup table. [Figure 14] Diagram (A) showing an example of pixel values when random numbers are added to the paper white portion of the reference image in Embodiment 1, and graph (B) representing G among the RGB values of the inspection image according to Embodiment 1. [Figure 15] Graph showing a comparison example between the inspection image and the reference image according to Embodiment 1. [Figure 16] Block diagram explaining the functional configuration of the inspection apparatus control unit of the inspection apparatus according to Embodiment 3. [Figure 17] Flowchart explaining the state determination process by the inspection apparatus according to Embodiment 3.

Modes for Carrying Out the Invention

[0013] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. Note that the following embodiments do not limit the invention according to the claims. Although a plurality of features are described in the embodiments, not all of these plurality of features are essential to the invention, and the plurality of features may be arbitrarily combined. Further, in the accompanying drawings, the same or similar configurations are denoted by the same reference numerals, and redundant descriptions are omitted.

[0014] FIG. 1 is a diagram showing a configuration example of an inspection system including an inspection apparatus according to Embodiment 1 of the present invention.

[0015] The image forming apparatus (printing apparatus) 100 processes various input data and prints onto a recording medium such as paper or a sheet to produce printed material. The inspection apparatus (image processing apparatus) 200 receives the printed material output from the image forming apparatus 100 and inspects its contents. The finisher 300 receives the printed material inspected by the inspection apparatus 200 and performs post-processing such as binding, stitching, or punching. The image forming apparatus 100 is connected to an external print server or client PC via a network. The inspection apparatus 200 is also connected to the image forming apparatus 100 on a one-to-one basis via a communication cable. The finisher 300 is also connected to the image forming apparatus 100 on a one-to-one basis via a different communication cable. Furthermore, the inspection apparatus 200 and the finisher 300 are also interconnected via another communication cable. Embodiment 1 shows an example of an inline inspection system that performs image formation, image inspection, and finishing in an integrated manner.

[0016] Figure 2 is a block diagram illustrating the hardware configuration of the image forming apparatus 100 according to Embodiment 1.

[0017] This image forming apparatus 100 includes a controller (control unit) 21, a printer unit 206, and a UI unit (operation unit) 23. The UI unit 23 includes various switches, LEDs, displays, etc., for operation.

[0018] Image data and document data transmitted to the image forming apparatus 100 are created by software applications such as printer drivers (not shown) on client PCs or print servers on the network. This image data and document data are transmitted to the image forming apparatus 100 as PDL data via the network (e.g., Local Area Network). In the image forming apparatus 100, the controller 21 receives the transmitted PDL data.

[0019] The controller 21 is connected to the UI unit 23 and the printer unit 206. It receives PDL data sent from a client PC or print server, converts it into print data that can be processed by the printer unit 206, and outputs the print data to the printer unit 206. The printer unit 206 prints the image based on the print data output from the controller 21. The printer unit 206 in Embodiment 1 is assumed to have an electrophotographic printing engine. However, the printing method is not limited to this, and for example, an inkjet method may also be used.

[0020] The UI unit 23 is operated by the user and is used to select various functions and give operation instructions. This UI unit 23 includes a display with a touch panel on its surface, and a keyboard with various keys such as a start key, stop key, and numeric keypad.

[0021] Next, we will describe the details of the controller 21.

[0022] The controller 21 includes a network interface unit 101, a CPU 102, RAM 103, ROM 104, an image processing unit 105, an engine interface unit 106, and an internal bus 107. The network interface unit 101 receives PDL data transmitted from a client PC or print server via the network. The CPU 102 uses programs and data stored in RAM 103 and ROM 104 to control the entire image forming apparatus 100 and execute the processing performed by the controller 21, as described later. RAM 103 has a work area used by the CPU 102 when executing various processes. ROM 104 stores programs and data for the CPU 102 to execute the various processes described later, as well as configuration data for the controller 21.

[0023] The image processing unit 105 performs image processing for printing on the PDL data received by the network I / F unit 101 according to the settings from the CPU 102, and generates printable data that can be output by the printer unit 206. In particular, the image processing unit 105 generates image data with multiple color components per pixel by rasterizing the received PDL data. Here, multiple color components refer to independent color components in a color space such as RGB (red, green, blue). The image data has 8 bits (256 gradations) for each color component for each pixel. That is, the image data is multi-level bitmap data that includes multi-level pixel data. In addition to the image data, the above rasterization also generates attribute data that shows the attributes of each pixel in the image data. This attribute data indicates what type of object the pixel belongs to, and is a value that indicates the type of object, such as text, line, graphic, image, or background. The image processing unit 105 generates print data by performing image processing such as color conversion from the RGB color space to the CMYK (cyan, magenta, yellow, black) color space and screen processing using the generated image data and attribute data. The engine I / F unit 106 is an interface that outputs the print data generated by the image processing unit 105 to the printer unit 206. The internal bus 107 is a system bus that connects the above-mentioned parts.

[0024] Figure 3 is a diagram illustrating the configuration of the printer unit 206 of the image forming apparatus 100 according to Embodiment 1.

[0025] The image forming apparatus 100 includes a scanner unit 301, a laser exposure unit 302, a photosensitive drum 303, an image formation unit 304, a fixing unit 305, a paper feeding / transport unit 306, and a printer control unit 308 that controls these units. The scanner unit 301 illuminates a document placed on the document table, optically reads the image on the document, and converts the image into an electrical signal to create image data.

[0026] The laser exposure unit 302 directs a light beam, such as laser light modulated according to the image data, onto a rotating polyhedron mirror (polygon mirror) 307 that rotates at a constant angular velocity, and irradiates the photosensitive drum 303 with the reflected light as scanning light. The image development unit 304 rotates the photosensitive drum 303, charges it with a charger, and develops the latent image formed on the photosensitive drum 303 by the laser exposure unit 302 with toner. This electrophotographic process is realized by having four development units (developing stations) that transfer the toner image to paper and collect any minute toner remaining on the photosensitive drum 303 that was not transferred.

[0027] The four developing units, arranged in the order of cyan (C), magenta (M), yellow (Y), and black (K), perform the image processing in the order of magenta, yellow, and black after a predetermined time has elapsed since the start of image processing at the cyan station.

[0028] The fixing unit 305 includes a combination of rollers and belts, and the heating roller has a built-in heat source such as a halogen heater, which melts the toner image on the paper onto which the toner image has been transferred by the image forming unit 304 using heat and pressure, thereby fixing it to the paper. However, when printing on thick paper, because the paper is thick and has poor thermal conductivity, the paper transport speed passing through the fixing unit 305 must be reduced to, for example, half the normal speed. Consequently, when printing on thick paper, the paper transport speed of each part other than the fixing unit 305 is also halved, so the printing speed of the image forming apparatus 100 itself is halved.

[0029] The paper feeding / transportation unit 306 is equipped with one or more paper storage compartments, such as paper cassettes or paper decks, and, in accordance with instructions from the printer control unit 308, separates one sheet of paper from among the multiple sheets stored in the paper storage compartment and transports it to the image formation unit 304. Furthermore, the paper onto which the toner image has been transferred by the image formation unit 304 is transported to the fuser unit 305. In this way, the paper is transported, and the toner images of each color are transferred by the aforementioned developing station, and finally a full-color toner image is formed on the paper. When forming an image on both sides of the paper, the transport path is controlled so that the paper that has passed through the fuser unit 305 is transported again to the image formation unit 304.

[0030] The printer control unit 308 communicates with the controller 21, which controls the entire image forming apparatus 100, and executes control according to its instructions. The printer control unit 308 also manages the status of each of the aforementioned parts, such as the scanner, laser exposure, image formation, fixing, and paper feeding / transport, and issues instructions to ensure that the entire system operates smoothly and in harmony.

[0031] Figure 4(A) is a schematic diagram illustrating the internal configuration of the inspection device 200 according to Embodiment 1.

[0032] The printed paper (printed material) from the image forming apparatus 100 is drawn into the inspection apparatus 200 by the paper feed roller 401. The printed material is then transported by the conveyor belt 402 and read by the inspection sensor 403 located above the conveyor belt 402. The inspection device control unit 405 uses the image data (scanned image (inspection image)) read by the inspection sensor 403 to perform inspection processing. The inspection device control unit 405 also controls the entire inspection apparatus 200. The inspection results are then sent to the subsequent finisher 300. The inspected printed material is then discharged by the paper discharge roller 404. Although not shown here, the inspection sensor 403 may also be configured to read from below the conveyor belt 402 to accommodate double-sided printed materials.

[0033] Figure 4(B) is a top view of the conveyor belt 402 as seen from the inspection sensor 403 side.

[0034] Here, the inspection sensor 403 is a line sensor that reads an image of the entire surface of the conveyed printed material 410 line by line, as shown in the figure. The image reading illumination device 411 illuminates the printed material when it is read by the inspection sensor 403. The skew detection illumination device 412 is used to detect whether the printed material 410 is skewed relative to the conveying direction as it is conveyed on the conveyor belt 402. The skew detection illumination device 412 detects skew by illuminating the conveyed printed material 410 from an oblique direction and reading the image of the shadow of the edge of the printed material. In Embodiment 1, the reading of the shadow image of the edge of the printed material is performed by the inspection sensor 403, but it may also be performed by a reading sensor other than the inspection sensor 403.

[0035] Figure 5 is a block diagram illustrating the functional configuration of the inspection device control unit 405 of the inspection device 200 according to Embodiment 1.

[0036] The control of the inspection device control unit 405 is entirely performed by the control unit 503. The control unit 503 has a CPU 515, which executes a program loaded into the memory unit 504 to perform various processes described later. The image input unit 501 receives scanned image data of the object to be inspected (hereinafter referred to as scanner image or inspection image) obtained by reading the inspection sensor 403. The CPU 515 stores this received scanned image in the memory unit 504. The communication unit 502 communicates with the controller 21 of the image forming apparatus 100. This communication involves receiving image data used for printing (reference image data) corresponding to the scanned image, and sending and receiving inspection control information. The CPU 515 also stores the received reference image data (hereinafter referred to as reference image) and inspection control information in the memory unit 504.

[0037] One type of inspection control information exchanged with the image forming apparatus 100 is synchronization information for matching scanned images with reference images, such as print job information, print quantity information, and page order information. Another type of information is inspection result information and control information that controls the operation of the image forming apparatus 100 accordingly. Synchronization information is necessary to synchronize the reference image and scanned image in cases where the order in which the scanned image and the reference image used to print the scanned image are received by the inspection control device 102 differs, such as in double-sided printing or printing multiple copies. Synchronization information is also necessary to synchronize the reference image and scanned image in cases where one reference image corresponds to multiple scanned images. The inspection control information exchanged with the finisher 300 is inspection result information and control information that controls the operation of the finisher 300 accordingly.

[0038] The inspection processing unit 513 is controlled by the CPU 515 of the control unit 503. Based on the synchronization information, which is one of the inspection control information exchanged with the image forming apparatus 100 mentioned above, the control unit 503 sequentially performs inspection processing on corresponding pairs of inspection images and reference images using the inspection processing unit 513. Details of the inspection processing unit 513 will be described later.

[0039] Once the inspection process is complete, the result is sent to the control unit 503 and displayed on the operation / display unit 505. If the result of this inspection determines that there is an abnormality, the operation / display unit 505 switches the control of the image forming apparatus 100 and the finisher 300 via the communication unit 502 in a manner predetermined by the user. For example, it may stop the image forming process by the image forming apparatus 100 and switch the output tray of the finisher 300 to the escape tray.

[0040] Next, I will explain the configuration of the inspection processing unit 513.

[0041] Oblique lineThe detection unit 506 is a module that detects the skew angle of the scanned image. As previously mentioned with reference to Figure 4(B), the scanned image is scanned in such a way that a shadow is cast at the edge of the printed material. This is so that the shadow at the edge, created when the printed material is illuminated by the skew detection illumination device 412 as it is pulled into the inspection device 200 and transported on the conveyor belt 402, can be scanned by the inspection sensor 403. This shadow is used to detect the skew angle of the printed material. line The angle is detected. line Based on the angle, correction processing is performed by the image deformation unit 509, which will be described later.

[0042] The color conversion unit 507 is a module that performs color conversion between the scanned image and the reference image. The reference image is image data rasterized in the CMYK color space by the image processing unit 105 of the image forming apparatus 100, and the scanned image is image data in the RGB color space read and acquired by the inspection sensor 403. The color conversion unit 507 converts the reference image to an RGB image. For example, the conversion may be performed using a CMYK_to_RGB lookup table (CMYK to RGB conversion table) as shown in Figure 13(C).

[0043] In this case, pixel data located on a grid point is converted to RGB color by referring to this conversion table, while pixel data not located on a grid point is interpolated from an adjacent grid point to obtain its RGB values.

[0044] Figure 13(A) shows an example of a reference image.

[0045] Region 1301 is the non-printed area (hereinafter referred to as white paper) in the reference image and represents a pixel region with a one-dimensional X axis. The X coordinates of this region 1301 are taken from 0 to N. Since region 1301 is white paper, all pixels in the shaded region 1301 have a CMYK value of (0,0,0,0). When converted by the color conversion unit 507 using the CMYK to RGB conversion table described above, the RGB value becomes (220,220,220). Note that since the RGB value is an 8-bit value from 0 to 255, the RGB value is converted to (220,220,220) to prevent saturation in the random number addition described later. Alternatively, the color conversion may be performed using a CMYK to RGB conversion table that takes into account the results of the random number assignment unit 511 described later.

[0046] The resolution conversion unit 508 is a module that converts the resolution of the scanned image and the reference image. The scanned image and the reference image may have different resolutions when they are input to the inspection device control unit 405. Also, the resolution used by each module of the inspection processing unit 513 may differ from the input resolution. In such cases, this module converts the resolution. For example, suppose the scanned image is 600 dpi for the main scan and 300 dpi for the sub-scan, and the reference image is 1200 dpi for the main scan and 1200 dpi for the sub-scan. If the resolution required by the inspection processing unit 513 is 300 dpi for both the main scan and sub-scan, then each image is scaled down so that both images are 300 dpi for both the main scan and sub-scan. The method of scaling can be any known method, taking into account the computational load and the required precision. For example, scaling using the SINC function is computationally intensive but yields high-precision scaling results. Scaling using the nearest neighbor method is computationally intensive but yields low-precision scaling results.

[0047] The image deformation unit 509 is a module that deforms the scanned image and the reference image. Between the scanned image and the reference image, there are factors such as paper expansion and contraction during printing and skew. line , skew during scanning line Geometric differences exist due to the above. The image deformation part 509 is oblique. lineThe geometric differences are corrected by deforming the image based on the information obtained from the detection unit 506 and the alignment unit 510, which will be described later. For example, geometric differences include linear transformations (rotation, scaling, shearing) and translation. These geometric differences can be expressed as affine transformations, and the affine transformation parameters are obliquely adjusted. line Correction can be performed by obtaining data from the detection unit 506 and the alignment unit 510. line The information obtained from the detection unit 506 is a parameter related to rotation (diagonal). line Only angle information is provided.

[0048] The alignment unit 510 is a module that performs alignment between a scanned image and a reference image. It is assumed that the scanned image and the reference image input to this module are images of the same resolution. Note that the higher the input resolution, the more accurate the alignment becomes, but the greater the computational load. By performing correction in the image deformation unit 509 based on the parameters obtained from the alignment, it is possible to obtain the inspection image and reference image to be used in the matching unit 512, which will be described later. Various alignment methods can be considered, but in Embodiment 1, in order to reduce the computational load, a method is used that performs alignment of the entire image using information from a part of the image rather than the entire image. The alignment according to Embodiment 1 includes three steps: selection of alignment patches, alignment of each patch, and estimation of affine transformation parameters. Each of these steps will be explained below.

[0049] First, let's explain the selection of alignment patches. Here, "patch" refers to a rectangular region within an image. In selecting alignment patches, several patches suitable for alignment are selected from the reference image. Patches with large corner features are considered suitable for alignment. Corner features are features where two distinct edges with different directions exist in a local neighborhood (the intersection of two edges). Corner features are features that represent the strength of these edge features. Various methods have been proposed based on differences in how "edge features" are modeled.

[0050] One known method for calculating corner features is the Harris corner detection method. The Harris corner detection method calculates a corner feature image from a horizontal differential image (horizontal edge feature image) and a vertical differential image (vertical edge feature image). This corner feature image represents the amount of the weaker of the two edges that make up the corner feature. Since both edges in a corner feature should be strong, the magnitude of the corner feature is expressed by whether the relatively weaker edge also has a strong edge amount.

[0051] Corner feature images are calculated from a reference image, and areas with large corner features are selected as patches suitable for alignment. Simply selecting areas with large corner features in order may result in patches being selected only from biased areas. In such cases, the number of areas without patches increases, and the image deformation information of those areas cannot be used, making it unsuitable for overall image alignment. Therefore, when selecting patches, not only the size of the corner features is considered, but also how the patches are distributed within the image. Specifically, even if the corner feature value of a candidate patch area is not large within the overall image, if the value is large within a local area of ​​the image, it will be selected as a patch. This makes it possible to distribute patches within the reference image. Parameters for patch selection include patch size and the number (or density) of patches. Larger patches and a higher number of patches improve alignment accuracy, but the computational load increases.

[0052] Next, we will explain patch-by-patch alignment. Patch-by-patch alignment involves aligning the alignment patch in the reference image selected in the previous step with the corresponding patch in the scanned image.

[0053] As a result of the alignment process, two types of information are obtained. The first is the center coordinates (refpX_i, reppY_i) of the alignment patch in the i-th reference image (i=1 to N, where N is the number of patches). The second is the position of those center coordinates in the scanned image (scanpX_i, scanpY_i). Any alignment method is acceptable as long as it is a method for estimating the shift amount that can obtain the relationship between (refpX_i, reppY_i) and (scanpX_i, scanpY_i). For example, one method could be to use FFT to place the alignment patch and the corresponding patch in frequency space, calculate their correlation there, and estimate the shift amount.

[0054] Finally, we will explain the estimation of affine transformation parameters. The affine transformation is a coordinate transformation method expressed by the equation shown in Figure 13(B).

[0055] In this equation, there are six types of affine transformation parameters: a, b, c, d, e, and f. Here, (x, y) corresponds to (refpX_i, refpY_i), and (x', y') corresponds to (scanpX_i, scanpY_i). The affine transformation parameters are estimated using this correspondence obtained from N patches. For example, the affine transformation parameters can be determined using the least squares method. Based on the affine transformation parameters thus obtained, the image deformation unit 509 deforms the reference image or the scanned image to create image data after alignment correction. In this way, a pair of reference image and inspection image to be used for matching in the matching unit 512 can be obtained.

[0056] The random number generator 511 is a module that assigns random numbers to the reference image. The random numbers assigned are used to adjust for the image difference between the scanned image and the reference image, as a difference exists between the scanned image and the reference image even if there are no abnormalities. This difference arises from the influence of the characteristics of the image forming apparatus, the characteristics of the scanner, etc. The characteristics of the image forming apparatus include color reproducibility, dot gain, and gamma characteristics. The characteristics of the scanner include color reproducibility, S / N ratio, and scanner MTF. The color conversion unit 507 removes the differences in color reproducibility of the image forming apparatus and scanner. Therefore, the random number generator 511 assigns random numbers to the reference image to remove differences in other noise components. As an example, we will explain the process of assigning the i-th horizontal and j-th vertical random number pattern to each region of the divided reference image, as shown in Figure 8, with a size of 9x9 vertically, as shown in Figure 7.

[0057] Figure 7 shows an example of a random number map according to Embodiment 1. Figure 8 shows an example of a reference image divided according to the embodiment.

[0058] As shown in Figure 8, a reference image with 45 pixels vertically (hereinafter, pixel = px) and 45 pixels horizontally is divided into multiple regions of the same size as the random number pattern. Here, the assignment of random numbers will be explained using region 802 in Figure 8 as an example. The image of region 802 is in a coordinate system with the top-left origin (0,0). Random numbers are added to this image by applying the random number pattern 701 from Figure 7(A). The random numbers added here are the random numbers for the coordinates (i,j) of random number pattern 701 that correspond to the pixel coordinate position. For example, if the pixel corresponding to coordinate (1,0) has RGB values ​​(100,101,102), the random number added to this will be the random number for the coordinates (i=1,j=0) of random number pattern 701. In the example in Figure 7(A), the random number for coordinate (1,0) is "1", so "1" is added to each of the RGB values. Similarly, by assigning random numbers to the entire reference image, random numbers can be assigned to each region of the divided reference image.

[0059] Similarly, by dividing a scanned image into multiple regions and performing the same process on all the divided regions of the scanned image, random numbers can be assigned to the entire scanned image.

[0060] Here, even a small change in the magnitude of the added random number can be effective. In Embodiment 1, the random number pattern was varied between 0 and 2, but if the pixel value can be calculated in decimal units during internal processing, it may be varied in decimal units between 0 and 1. Here, it is necessary to add noise of a magnitude or brightness that will not be detected as an anomaly.

[0061] Although Embodiment 1 described an example of adding a random number greater than or equal to 0, the present invention is not limited to this. For example, positive and negative values ​​centered around 0 may be used, or random numbers may be subtracted instead of added.

[0062] While there are no specific limitations on the random number patterns, it is desirable that the window size used for block matching by nearest neighbor search, described later, includes both noise and noise-free random number patterns. (Random numbers in Figure 7(A)) pattern Taking 701 as an example, each pixel of interest contains both a "0" (no noise) and a "1" or "2" (no noise) within the window. Note that the random number pattern is not limited to this; for example, each RGB value could have its own random number pattern.

[0063] Here, we will explain an example of random number generation using the reference image in Figure 13(A) as an example.

[0064] The RGB values ​​of region 1301 are as converted by the color conversion unit 507 (220,220,220). This one-dimensional region is represented by coordinates (X,0:X is 0-8), and random number pattern 701 is added to the pixels corresponding to each coordinate. The result is shown in graph 1401 in Figure 14(A). In graph 1401, the vertical axis represents the RGB pixel values, and the horizontal axis represents the X coordinate. Graph 1401 shows an example where the random numbers from the first row (j=0) of pattern 701 in Figure 7(A) are added. In the reference image, the white pixels always had a constant value, but it can be seen that they fluctuate when random numbers are added.

[0065] Returning to Figure 5, the comparison unit 512 is a module that compares the inspection image with the reference image. The inspection image and reference image input to this module are image data of the same resolution. In addition, the alignment unit 510 is used to enable image comparison. in It is assumed that the reference image or inspection image has been corrected by the image deformation unit 509 based on the information obtained.

[0066] The matching unit 512 creates a matching image using the reference image and the inspection image to which random numbers have been assigned by the random number generation unit 511. Here, when creating the matching image, by further aligning the position with higher precision based on the information obtained by the alignment unit 510, high-precision anomaly detection becomes possible. In Embodiment 1, high-precision alignment is achieved by block matching using nearest neighbor search. The matching process is performed based on parameters notified from the operation unit / display unit 505. Details of the matching process will be described later. Note that high-precision alignment is not limited to block matching. It is desirable to perform detailed alignment in a narrow area from the information obtained by the alignment unit 510, so alignment of feature points in a local area is also acceptable.

[0067] The operation / display unit 505 is a touchscreen user interface that accepts settings for processing in the inspection processing unit 513 from the user. For example, the operation / display unit 505 displays a settings screen, such as the one shown in Figure 10, and accepts settings for image processing by the inspection processing unit 513 from the user.

[0068] Figure 10 shows an example of a UI screen displayed on the operation / display unit 505 of the inspection device 200 according to Embodiment 1.

[0069] Here, there are five user-adjustable inspection settings: Settings 1 through 5. For example, when Setting 1 is set, the matching unit 512 determines an abnormality if the color difference of stains, scratches, etc., determined by the inspection of the inspection image is "5" or greater. On the other hand, when Setting 5 is set, the matching unit 512 determines an abnormality if the color difference of stains, scratches, etc., determined by the inspection of the inspection image is "50" or greater. In the example in Figure 10, the smaller the setting number of the inspection setting, the more likely the matching unit 512 is to determine an abnormality even for slight color differences of stains, scratches, etc. In this way, the user can select one of the setting buttons from Settings 1 through 5 and press the "Enable" button to enable the matching unit 512. And A threshold value for judgment can be set. The calculation of color difference will be explained in detail in the matching judgment process described later. In the example in Figure 10, each setting value is pre-associated with a color difference parameter. Therefore, the color difference parameter corresponding to the setting value selected by the user is notified from the operation unit / display unit 505 to the matching unit 512.

[0070] In Embodiment 1, the color difference of the detected dirt and scratches was adjusted based on the setting values ​​shown in Figure 10, but the invention is not limited to this, and for example, the size of the detected dirt and scratches may be set. For example, the size may be set within a range of 0.1 mm to 3 mm.

[0071] Next, the inspection process using the inspection device 200 according to Embodiment 1 will be described.

[0072] Figure 6 is a flowchart illustrating the inspection process by the inspection device 200 according to Embodiment 1. The process shown in this flowchart is realized by the CPU 515 of the control unit 503 executing a program stored in the memory unit 504. At this time, the CPU 515 performs the inspection shown in Figure 5. product Each processing unit functions as a processing unit of the processing unit 513, and the processed results are stored in the memory unit 504 and used in subsequent processing.

[0073] First, the S601 has a CPU of 515 teeth Next, pre-processing for inspection is performed. At this time, the CPU 515 uses the inspection control information received from the image forming apparatus 100, which is stored in the memory unit 504 via the communication unit 502, to select a pair of scanned images and reference images to be processed. Then the CPU 515 skews the scanned image. line The detection unit 506 processes the scanned image to obtain tilt information. Based on this tilt information, the image deformation unit 509 performs correction processing on the scanned image. In parallel with this, the aforementioned reference image generation process is performed, and the reference image is processed by the color conversion unit 507 to create an image suitable for inspection.

[0074] Next, the process proceeds to S602, where the CPU 515 performs alignment using the scanned image and reference image obtained in S601. At this time, the CPU 515 converts the scanned image and reference image to a predetermined resolution (for example, 300 dpi × 300 dpi) using the resolution conversion unit 508. Then, the scan image and reference image with the predetermined resolution are processed by the alignment unit 510 to obtain affine transformation parameters. The CPU 515 then uses the affine transformation parameters obtained from the alignment unit 510 to perform correction processing on the reference image using the image deformation unit 509, making the coordinate system of the reference image the same as that of the scanned image, and thus making it an image that can be used for matching.

[0075] Then, proceeding to S603, the CPU 515 performs a comparison / judgment process using the inspection image (scan image) obtained in S802 and the reference image. At this time, the CPU 515 functions as a comparison unit 512 and compares the inspection image with the reference image. Then, proceeding to S604, the CPU 515 displays the result of the comparison process in S603 on the operation unit / display unit 505. Here, simply displaying the final judgment result image would make it difficult to understand what kind of image defect there was, so the final judgment result image is superimposed on the inspection image and displayed on the operation unit / display unit 505. Any synthesis method is acceptable as long as it makes it easy to understand the location of the image defect. For example, the part judged as "1" (the abnormal part) in the final judgment result image could be made red and superimposed on the inspection image.

[0076] Next, the matching process will be explained using the flowchart shown in Figure 11.

[0077] Figure 11 is a flowchart illustrating the verification process performed by the inspection device 200 according to Embodiment 1. The process shown in this flowchart is realized when the CPU 515 of the control unit 503 executes a program stored in the memory unit 504. At this time, the CPU 515 functions as the verification unit 512 shown in Figure 5.

[0078] First, in S1101, the matching unit 512 performs high-precision alignment of the reference image and the inspection image using nearest neighbor search. As mentioned above, the image deformation unit 509 performs alignment of the reference image and the inspection image, but in order to detect anomalies with high precision, it is necessary to perform further local alignment. Therefore, in Embodiment 1, high-precision alignment is performed using nearest neighbor search with block matching. A schematic diagram of the nearest neighbor search is shown here.

[0079] Figure 12 is a schematic diagram of the nearest neighbor search according to Embodiment 1.

[0080] Figure 12(A) shows an example of an inspection image, and Figure 12(B) shows an example of a reference image. Area 1201 (a predetermined region) of the reference image indicates the search area of ​​the reference image. The search window 1202 of the inspection image targets the 8 neighboring pixels of pixel 1203. Within area 1201 of the reference image, a search window 1204 corresponding to this search window 1202 is set. Then, this search window 1204 is moved 4 pixels to the right within area 1201 (x+4), and when this search window 1204 reaches the right edge of area 1201, the x coordinate is returned to its original value and the y coordinate is subtracted by 4, moving the search window 1204 directly down to the position of the initial search window 1204. This operation is repeated to sequentially search within area 1201 for the reference image region with the smallest difference between the search window 1202 of the inspection image and the reference image region. The difference Δarea to be obtained here is the sum of the target regions of the color difference ΔRGB shown in the following equation (1).

[0081] ΔRGB = √(R 2 +G2 +B 2 ) … Formula (1) Here, we will describe the difference between the case where a random number is generated by the random number generation unit 511 and the case where a random number is not generated.

[0082] In Figure 13(A), region 1301 of the reference image is described as white in the color conversion unit 507 and as RGB(220,220,220). When block matching is performed on this region 1301 using nearest neighbor search, no superiority or inferiority is determined for pixels in the white region alone. On the other hand, when random numbers are added, the maximum and minimum values ​​of Δarea change when viewed in each search window region. In block matching using nearest neighbor search, the part with the minimum value is selected in the search window region. Therefore, when random numbers are added to the reference image, alignment can be performed with higher accuracy.

[0083] Next, proceeding to S1102, the matching unit 512 calculates a difference image between the reference image and the inspection image, which were realigned in S1101. In Embodiment 1, the difference in G of RGB, ΔG, is used as the color difference for detecting abnormalities, and this example of calculating the color difference will be explained. In Embodiment 1, the difference in G of RGB values, ΔG, was used as the color difference, but abnormalities may also be detected using ΔRGB as the color difference.

[0084] Furthermore, for grayscale images, a simple absolute difference can be used, or a function that calculates the absolute difference while considering gamma characteristics can be used.

[0085] Figure 14(B) shows graph 1402, which represents the G value among the RGB values ​​of the inspection image. The vertical axis of this graph represents the pixel value of G, and the horizontal axis represents the X coordinate. Note that the X coordinates described here are those of a one-dimensional region of the inspection image that is in the same position as the reference image in Figure 13(A), and are assumed to contain no abnormalities but to be mixed with noise.

[0086] Figure 15 is a diagram showing graphs of reference images and inspection images according to Embodiment 1.

[0087] In Figure 15, the solid line represents G in the same inspection image as in Figure 14(B), and the dashed line represents the reference image. In Figure 15, the vertical axis of the graph represents the pixel value of G, and the horizontal axis represents the X coordinate. Figure 15(A) shows the case where no noise is added to the reference image. Figure 15(B) shows a comparison between the result of adding a random number pattern 701 to the reference image as shown in Figure 14(A) and the graph 1402 showing G in the inspection image in Figure 14(B). Figure 15(C) shows a comparative example of the inspection image and the reference image when alignment is performed by nearest neighbor search using the matching unit 512.

[0088] In Figure 15(A), the ΔG values ​​at X coordinate positions 4 and 5 are separated by 5 or more. In this case, as shown in Figure 10 above, if the inspection setting set in the operation / display unit 505 is setting 1, a ΔG separation of "5" or more is considered an abnormality, resulting in a false detection.

[0089] Similarly to Figure 15(A), in the case of Figure 15(B), the difference between X coordinate positions 4 and 5 is 5 or more, so a false detection will occur in setting 1.

[0090] In contrast, in Figure 15(C), the largest difference is "4" at X coordinate position 2, so in this case, even with setting 1 as described above, it is not detected as an anomaly.

[0091] As explained above, the above processing and the processing of the random number generation unit 511 would normally cause false detection due to noise contained in the inspection image. However, by performing local alignment between the inspection image and the reference image using nearest neighbor search while random numbers are also assigned to the reference image, the matching judgment process can be performed with the noise contained in the inspection image canceled out, thereby suppressing false detection.

[0092] Furthermore, the random numbers to be added may be changed according to the inspection settings based on the parameters notified from the operation unit / display unit 505. Specifically, a table may be provided that increases the amount of random numbers (noise) added when the inspection setting is high (large) and decreases the amount of random numbers (noise) added when the inspection setting is low (small), and this table may be switched according to the inspection setting.

[0093] Figure 7(B) shows an example of a random number pattern 702 when the check setting is increased by one level. As the check setting increases in this way, similarly disorder Switch to a table with a larger number (amount of noise). This is because varying the value of the assigned random number within a range of magnitude and brightness that will not be detected as an anomaly further enhances the effect of suppressing false positives.

[0094] (Modified version of Embodiment 1) The following describes image processing related to a modified embodiment 1 of the present invention.

[0095] In the above-described embodiment 1, the reference image was a digital image used for printing, and noise components in the inspection image and the reference image were canceled out to improve the accuracy of the matching. However, some inspection systems not only make the reference image a digital image, but also print and scan it in the same way as the inspection image, and use the scanned image that the user has determined to be free of abnormalities as the reference image.

[0096] When a scanned image is used as the reference image, the scanned image contains noise such as surface texture and transparency variations inherent in the paper, as well as the signal-to-noise ratio of the scanner. Therefore, there is no need to assign random numbers to it. Accordingly, when using a scanned image as the reference image, the random number generation unit 511 switches to not assigning random numbers.

[0097] As explained above, with this modification, whether the reference image is a digital image used for printing, or a scanned image obtained by scanning a printed document, similar to the inspection image, the matching process can be performed while canceling out noise components. Therefore, it is possible to suppress the false detection of abnormalities in the inspection image.

[0098] [Embodiment 2] The image processing according to Embodiment 2 of the present invention will be described below.

[0099] In the above-described embodiment 1, a method was explained in which noise components contained in the inspection image are canceled out by adding random numbers to the reference image, thereby enabling suitable matching and determination.

[0100] In Embodiment 1, when assigning random numbers, the random numbers were assigned according to the random number map in Figure 7, regardless of the pixel value of the pixel of interest. However, the noise contained in the inspection image is greater in the white and highlight areas of the paper than in the dark areas. This is because the inspection sensor 403 is an RGB light receiving element, and therefore has higher sensitivity in the highlight areas.

[0101] Therefore, in Embodiment 2, the value of the random number assigned is changed according to the pixel value. This makes it possible to assign more suitable random numbers to each part, such as dark areas, white areas, and highlight areas, which leads to the suppression of false detections. In the following, only the differences from Embodiment 1 will be described. The system configuration of Embodiment 2 and the hardware configuration of the image forming apparatus 100 and inspection apparatus 200 are the same as those of Embodiment 1 described above, so their explanation will be omitted.

[0102] The random number generator 511 according to Embodiment 2 references the pixel values ​​of a reference image and multiplies the random number to be generated by a coefficient corresponding to those pixel values, thereby achieving the generation of a suitable random number.

[0103] Similar to Embodiment 1, the explanation will be given using the image region 802 of the reference image in Figure 8. When the coordinate (2,0) has RGB values ​​(0,96,192), the random number pattern i=2, j=0 is added. Here, the following equation (2) shows an example of calculating R', in which the coefficient D for the dark area, the coefficient H for the white area, and the random value rand of the random number pattern (i,j) are assigned to R in the RGB value of the coordinate (2,0). In equation (2) below, the closer to the white area, the larger the random number added becomes, and in the dark area, it becomes smaller. This makes it possible to assign large random numbers to pixels in bright areas that are susceptible to noise, such as white areas and highlights, and small random numbers to pixels that are less susceptible to noise, such as dark areas.

[0104] R'=(((HD)*R)÷255+D)*rand+R…(Formula 2) From the above formula, when R=0, D=1, and H=5, rand is "2" and R' is 10. By calculating G and B similarly, it is possible to assign suitable random numbers that refer to the pixel values ​​of the reference image.

[0105] In addition, while random numbers were generated using a linear arithmetic formula in this invention, the present invention is not limited to this. Random numbers may also be generated by referring to a lookup table corresponding to the pixel values.

[0106] As described above, according to Embodiment 2, by changing the value of the random number assigned according to the pixel value of the reference image, it is possible to assign more suitable random numbers to each part, such as dark areas, white areas, and highlight areas, which has the effect of suppressing false detections.

[0107] [Embodiment 3] The following describes the process according to Embodiment 3 of the present invention.

[0108] In Embodiment 1 described above, a method was explained for suitably performing matching and judgment by canceling out noise components contained in the inspection image by assigning random numbers to the reference image. In Embodiment 1, when assigning random numbers, pre-set random numbers were assigned. However, the amount and distribution of noise change due to individual differences in the print output paper 410 and the inspection sensor 403. Therefore, Embodiment 3 will be explained as an example in which random numbers are assigned that take into account the individual differences in the print output paper 410 and the inspection sensor 403. Note that only the differences from Embodiment 1 will be explained below. The system configuration and the hardware configuration of the image forming apparatus 100 and inspection apparatus 200 related to Embodiment 3 are the same as in Embodiment 1 described above, so their explanation will be omitted.

[0109] Figure 16 is a block diagram illustrating the functional configuration of the inspection device control unit of the inspection device according to Embodiment 3. In Figure 16, parts identical to those in Figure 5 are indicated by the same reference numerals, and their descriptions are omitted.

[0110] The inspection processing unit 513 according to Embodiment 3 has a state determination unit 514 for determining the state of the print output paper 410 and the inspection sensor 403. The state determination unit 514 performs the printing process without printing anything on the print output paper 410 and measures the amount of noise by reading it with the inspection sensor 403. The inspection sensor 403 reads the four corner areas of the print output paper 410 as shown in Figure 9 and sends the values ​​to the state determination unit 514.

[0111] Figure 9 is a diagram illustrating the four corner regions of the print output paper 410 in Embodiment 3.

[0112] Figure 17 is a flowchart illustrating the state determination process by the inspection device 200 according to Embodiment 3. The process shown in this flowchart is realized by the CPU 515 of the control unit 503 executing a program stored in the memory unit 504. At this time, the CPU 515 functions as the state determination unit 514 shown in Figure 16.

[0113] First, in S1701, the state determination unit 514 removes the maximum and minimum values ​​of the RGB noise levels in each of the four corner regions of the printed output paper 410, as read by the inspection sensor 403. Next, in S1702, the state determination unit 514 determines whether the process in S1701 has been completed in all regions. If it has not been completed, S1701 is repeated; if it has been completed, the process proceeds to S1703. In S1703, the state determination unit 514 calculates the variance value of the noise level for all four corner regions. Here, we will explain using R from the RGB values ​​as an example. Variance value S 2 This can be calculated using the following equation (3), where n is the total number of pixels obtained by removing the maximum and minimum values ​​of R in S1701, and the individual pixel values ​​xi and the average value of the pixel values ​​xi.

[0114] S 2 =(1 / n)Σ{xi-(mean value of xi)} 2 ...Formula (3) Here, Σ represents the sum from n=1 to n=n.

[0115] Similarly, by calculating the values ​​for G and B, we can calculate the variance of all RGB values.

[0116] In Embodiment 3, the area read by the inspection sensor was set to the four corners to consider processing speed, but the present invention is not limited to this. Alternatively, a few percent of the RGB values ​​from the maximum and a few percent from the minimum values ​​could be removed from the entire sheet of paper, and the variance of the remaining RGB values ​​could be calculated.

[0117] The random number generator 511 changes the random number to be generated based on the variance value calculated by the state determination unit 514. Here, the random number map used may be switched based on the variance value, or the standard deviation and random number map may be generated. For example, if the threshold is "5" and the calculated variance value exceeds the threshold, the random number in Figure 7(A) pattern From 701, the random numbers assigned will increase as shown in Figure 7(B). pattern You could also switch to 702.

[0118] As described above, according to Embodiment 3, it is possible to adjust the amount of noise that takes into account individual differences in the print output paper 410 and the inspection sensor 403, enabling more accurate anomaly detection.

[0119] Furthermore, while Embodiment 3 described an example where the random numbers assigned were switched based on individual differences in the print output paper 410 and the inspection sensor 403, the present invention is not limited to this. The random numbers may also be changed by reading the CMYK gradation, taking into account the characteristics and status of the scanner and printer of the image forming apparatus 100.

[0120] (Other embodiments) The present invention can also be realized by supplying a program that implements one or more of the functions of the above-described embodiments to a system or device via a network or storage medium, and by having one or more processors in the computer of that system or device read and execute the program. It can also be realized by a circuit (e.g., an ASIC) that implements one or more functions.

[0121] [Item 1] An inspection device for inspecting images formed on a recording medium by a printing device, A storage means for storing the image data used to form the image as a reference image on the recording medium, An acquisition means for acquiring image data of an object to be inspected formed on the aforementioned recording medium, A means for adding noise components to the aforementioned reference image, A positioning means for aligning the reference image to which the noise component has been added by the aforementioned adding means with the image data to be inspected, A matching means that performs a matching process between a reference image aligned by the alignment means and the image data of the object to be inspected, An inspection device characterized by having the following features.

[0122] [Item 2] The inspection apparatus according to item 1, wherein the noise component is a random number, and the application means divides the reference image into a plurality of regions and applies a noise component pattern to each of the plurality of regions to apply the noise component.

[0123] [Item 3] The inspection apparatus according to item 1 or 2, characterized in that the assigning means assigns the noise component based on the pixel values ​​of the reference image.

[0124] [Item 4] The inspection apparatus according to item 3, characterized in that the imparting means imparts a larger noise component when the pixel value of the reference image indicates a bright area than when it indicates a dark area.

[0125] [Item 5] The inspection apparatus according to any one of items 1 to 4, characterized in that the acquisition means acquires image data of the object to be inspected by optically reading the image of the object to be inspected formed on the recording medium.

[0126] [Item 6] The inspection apparatus according to item 5, characterized in that the imparting means imparts the noise component in order to reduce the difference between the reference image and the image data to be inspected, which is caused by the noise component contained in the image data to be inspected.

[0127] [Item 7] The system further comprises means for determining the variance of pixel values ​​obtained by optically reading a recording medium in which no image has been formed, The inspection apparatus according to any one of items 1 to 6, characterized in that the imparting means modifies the noise component based on the variance.

[0128] [Item 8] The inspection apparatus according to any one of items 1 to 7, further comprising setting means for setting a threshold for determining whether or not there is an abnormality in the image based on the matching process in the matching means.

[0129] [Item 9] The inspection apparatus according to item 8, characterized in that the assigning means changes the noise component based on the threshold set by the setting means.

[0130] [Item 10] The inspection apparatus according to item 9, characterized in that the imparting means increases the noise component when the threshold increases and decreases the noise component when the threshold decreases.

[0131] [Item 11] The inspection apparatus according to item 5, further comprising a color conversion means for converting the color space of the reference image to the color space of the image data to be inspected.

[0132] [Item 12] The inspection apparatus according to any one of items 1 to 11, characterized in that the alignment means performs a nearest neighbor search using block matching between a search window set in the image data to be inspected and a predetermined region of a reference image to which the noise component has been added, and performs the alignment by selecting the block with the smallest difference in pixel values.

[0133] [Item 13] The inspection apparatus according to any one of items 1 to 12, further comprising a resolution conversion means for converting the resolution of the reference image or the image data to be inspected in order to make the resolution of the reference image and the image data to be inspected the same.

[0134] [Item 14] An inspection device for inspecting images formed on a recording medium by a printing device, A storage means for storing image data obtained by scanning the image formed on the aforementioned recording medium as a reference image, An acquisition means for acquiring image data of an object to be inspected by scanning an image of the object to be inspected formed on the recording medium, Alignment means for aligning the reference image with the image data of the object to be inspected, A matching means that performs a matching process between a reference image aligned by the alignment means and the image data of the object to be inspected, An inspection device characterized by having the following features.

[0135] [Item 15] The inspection apparatus according to item 13, characterized in that the alignment means performs a nearest neighbor search using block matching between a search window set in the image data to be inspected and a predetermined area of ​​the reference image, and performs the alignment by selecting the block with the smallest difference in pixel values.

[0136] [Item 16] The inspection apparatus according to item 13 or 14, further comprising setting means for setting a threshold for determining whether or not there is an abnormality in the image based on the matching process in the matching means.

[0137] [Item 17] The inspection apparatus according to any one of items 14 to 16, further comprising a resolution conversion means for converting the resolution of the reference image or the image data to be inspected in order to make the resolution of the reference image and the image data to be inspected the same.

[0138] [Item 18] A control method for controlling an inspection device that inspects an image formed on a recording medium by a printing device, A storage step of storing the image data used to form the image as a reference image in the recording medium, An acquisition step of acquiring image data of the object to be inspected formed on the recording medium, The process of adding noise components to the aforementioned reference image, A positioning step is performed to align the reference image to which the noise component has been added by the aforementioned addition step with the image data to be inspected. A matching step which performs a matching process between the reference image aligned by the alignment step and the image data of the object to be inspected, A control method characterized by having the following features.

[0139] [Item 19] A control method for controlling an inspection device that inspects an image formed on a recording medium by a printing device, A storage step of storing image data obtained by scanning the image formed on the recording medium as a reference image, An acquisition step of acquiring image data of the object to be inspected by scanning the image of the object to be inspected formed on the recording medium, A positioning step in which the reference image and the image data of the object to be inspected are aligned, A matching step which performs a matching process between the reference image aligned by the alignment step and the image data of the object to be inspected, A control method characterized by having the following features.

[0140] [Item 20] A program for causing a computer to function as one of the means of an inspection apparatus described in any one of items 1 through 17.

[0141] The present invention is not limited to the embodiments described above, and various modifications and variations are possible without departing from the spirit and scope of the invention. Accordingly, the following claims are attached to make the scope of the invention public. [Explanation of Symbols]

[0142] 100…Image forming apparatus, 200…Inspection apparatus, 300…Finisher, 502…Communication unit, 503…Control unit, 504…Memory unit, 506…Skew detection unit, 507…Color conversion unit, 508…Resolution conversion unit, 510…Alignment unit, 511…Random number generation unit, 512…Verification unit, 515…CPU, 514…State determination unit

Claims

1. An inspection device for inspecting images formed on a recording medium by a printing device, A storage means for storing the image data used to form the image as a reference image on the recording medium, An acquisition means for acquiring image data of an object to be inspected formed on the aforementioned recording medium, A means for adding noise components to the aforementioned reference image, A positioning means for aligning the reference image to which the noise component has been added by the aforementioned adding means with the image data to be inspected, A matching means that performs a matching process between a reference image aligned by the alignment means and the image data of the object to be inspected, The system includes means for determining the variance of pixel values ​​obtained by optically reading a recording medium in which no image has been formed, The inspection apparatus is characterized in that the imparting means modifies the noise component based on the variance.

2. The inspection apparatus according to claim 1, wherein the noise component is a random number, and the assigning means divides the reference image into a plurality of regions and assigns the random number by applying a random number pattern to each of the plurality of regions.

3. The inspection apparatus according to claim 2, characterized in that the assigning means assigns the random numbers based on the pixel values ​​of the reference image.

4. The inspection apparatus according to claim 3, characterized in that the assigning means assigns a larger random number when the pixel value of the reference image indicates a bright area than when it indicates a dark area.

5. The inspection apparatus according to claim 1, characterized in that the acquisition means acquires image data of the object to be inspected by optically reading the image of the object to be inspected formed on the recording medium.

6. The inspection apparatus according to claim 5, characterized in that the imparting means imparts the noise component in order to reduce the difference between the reference image and the image data to be inspected, which is caused by the noise component contained in the image data to be inspected.

7. The inspection apparatus according to claim 1, further comprising setting means for setting a threshold for determining whether or not there is an abnormality in the image based on the matching process in the matching means.

8. The inspection apparatus according to claim 7, characterized in that the assigning means modifies the noise component based on the threshold set by the setting means.

9. The inspection apparatus according to claim 8, characterized in that the imparting means increases the noise component when the threshold increases and decreases the noise component when the threshold decreases.

10. The inspection apparatus according to claim 5, further comprising a color conversion means for converting the color space of the reference image to the color space of the image data to be inspected.

11. The inspection apparatus according to claim 1, characterized in that the alignment means performs a nearest neighbor search using block matching between a search window set in the image data to be inspected and a predetermined region of a reference image to which the noise component has been added, and performs the alignment by selecting the block with the smallest difference in pixel values.

12. The inspection apparatus according to claim 1, further comprising a resolution conversion means for converting the resolution of the reference image or the image data of the object to be inspected in order to make the resolution of the reference image and the image data of the object to be inspected the same.

13. An inspection device for inspecting an image formed on a recording medium by a printing device, A storage means for storing the image data used to form the image as a reference image on the recording medium, An acquisition means for acquiring image data of an object to be inspected formed on the aforementioned recording medium, A means for adding noise components to the aforementioned reference image, A positioning means for aligning the reference image to which the noise component has been added by the aforementioned adding means with the image data to be inspected, A matching means that performs a matching process between a reference image aligned by the alignment means and the image data of the object to be inspected, The matching means includes a setting means for setting a threshold for determining whether or not there is an abnormality in the image based on the matching process in the matching means, The inspection apparatus is characterized in that the assigning means modifies the noise component based on the threshold set by the setting means.

14. A control method for controlling an inspection device that inspects an image formed on a recording medium by a printing device, A storage step of storing the image data used to form the image as a reference image in the recording medium, An acquisition step of acquiring image data of the object to be inspected formed on the recording medium, The process of adding noise components to the aforementioned reference image, A positioning step is performed to align the reference image to which the noise component has been added by the aforementioned addition step with the image data to be inspected. A matching step which performs a matching process between the reference image aligned by the alignment step and the image data of the object to be inspected, The process includes a step of determining the variance of pixel values ​​obtained by optically reading a recording medium in which no image has been formed, The aforementioned addition step is a control method characterized by changing the noise component based on the variance.

Images