Image inspection apparatus, image forming apparatus, and image inspection method

Abstract

To provide an image inspection device and an image inspection method capable of inspecting an inspection object at high speed and with high accuracy.SOLUTION: An image inspection device according to an aspect of the present invention is an image inspection device that inspects a defect of an inspection target, including an imaging unit that captures an inspection image of the inspection target, and a processing unit that processes the inspection image captured by the imaging unit, in which the processing unit generates a sensitivity map image that is an image having the same size as the inspection image and a non-defective product image and in which a detection threshold is set for each predetermined detection area, and detects a defective area using the sensitivity map image.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to an image inspection device, an image forming device, and an image inspection method. [Background technology]

[0002] For example, Patent Document 1 discloses a technique for extracting defect candidate regions by comparing the brightness values ​​of pixels in a divided image with those of surrounding pixels in a reduced resolution in order to inspect an object for abnormalities at high speed and with high sensitivity. Patent Document 2 discloses a technique for increasing the inspection accuracy of an object such as a printed matter by setting different defect detection thresholds for regions with large changes in pixel values, such as patterns and edges, and regions with small changes in pixel values, such as the background. Summary of the Invention [Problem to be solved by the invention]

[0003] However, for example, when the technology of Patent Document 1 is applied to the inspection of an object that includes areas with large changes in pixel values ​​and areas with small changes in pixel values, the defect detection threshold cannot be set appropriately in the low-resolution image, which may result in low inspection accuracy.

[0004] An object of the present invention is to provide an image inspection device and an image inspection method that are capable of inspecting an object to be inspected at high speed and with high accuracy. [Means for solving the problem]

[0005] An image inspection device according to one aspect of the present invention is an image inspection device for inspecting defects in an object to be inspected, the image inspection device comprising: an imaging unit that captures an inspection image of the object to be inspected; and a processing unit that processes the inspection image captured by the imaging unit, the processing unit comprising: an inspection image divided image generation unit that generates an inspection image divided image using a first inspection image processed image obtained by averaging pixel brightness values ​​in each of a plurality of processing regions obtained by dividing the inspection image according to a predetermined first inspection image division condition; a good product image divided image generation unit that generates a good product image divided image using a first good product image processed image obtained by averaging pixel brightness values ​​in each of a plurality of processing regions obtained by dividing a good product image to be compared with the inspection image according to the predetermined first inspection image division condition; a sensitivity map image generation unit that generates a sensitivity map image of the same size as the inspection image and the good product image, in which a detection threshold is set for each predetermined detection region; and a sensitivity map image generation unit that divides the sensitivity map image into a plurality of regions according to the first inspection image division condition and processes the first inspection image threshold for each region according to the detection threshold. a first inspection image sensitivity image generating unit that generates a first inspection image sensitivity image obtained by dividing the sensitivity map image into a plurality of regions in accordance with the first non-defective image division conditions and processing the first non-defective image threshold of each region in accordance with the detection threshold; and a first non-defective image sensitivity image generating unit that compares the brightness values ​​of pixels in a region of interest for an inspection image among a plurality of processing regions included in the divided image for an inspection image with the brightness values ​​of pixels in the processing regions around the region of interest for an inspection image, and generates the first inspection image sensitivity image. the inspection image defect candidate extraction unit extracting a defect candidate area for an inspection image based on a first threshold for good image; the good image defect candidate extraction unit comparing the brightness values ​​of pixels in a region of interest for a good image among a plurality of processing regions included in the divided image for good image with the brightness values ​​of pixels in the processing regions around the region of interest for a good image, and extracting a defect candidate area for a good image based on the first threshold for good image; and the defect detection unit comparing the inspection image defect candidate area with the defect candidate area for a good image to detect a defect area. [Effects of the Invention]

[0006] According to the present invention, it is possible to provide an image inspection device and an image inspection method that are capable of inspecting an object to be inspected at high speed and with high accuracy. [Brief explanation of the drawings]

[0007] [Figure 1] 1 is a diagram illustrating an example of the overall configuration of an image forming apparatus including an image inspection device according to an embodiment. [Figure 2] 1 is a block diagram illustrating an example of the hardware configuration of an image forming apparatus including an image inspection apparatus according to an embodiment. [Figure 3] 1 is a diagram illustrating an example of the configuration of an image inspection device according to an embodiment. [Figure 4] 2 is a block diagram showing an example of the functional configuration of a processing unit included in the image inspection device according to the first embodiment. FIG. [Figure 5] FIG. 3 is a flowchart showing the operation of the image inspection device according to the first embodiment. [Figure 6] FIG. 10 is a diagram showing an example of an inspection image. [Figure 7] FIG. 10 is a diagram showing an example of a processed image for a first inspection image. [Figure 8] FIG. 8 is a diagram showing an example of an inspection image defect candidate area image obtained from the first inspection image processed image of FIG. 7. [Figure 9] FIG. 10 is a diagram showing an example of a non-defective product image. [Figure 10] FIG. 10 is a diagram showing an example of a first processed image for a non-defective product image. [Figure 11] 11 is a diagram showing an example of a defect candidate region for a non-defective image obtained from the first processed image for a non-defective image of FIG. 10. FIG. [Figure 12] 10A and 10B are diagrams illustrating a sensitivity map image and a sensitivity image for a first inspection image. [Figure 13] 10A and 10B are diagrams illustrating averaging processing of the sensitivity image for the first inspection image. [Figure 14] FIG. 10 is a diagram showing maximum value processing of the sensitivity image for the first inspection image. [Figure 15] FIG. 10 is a diagram showing minimum value processing of the sensitivity image for the first inspection image. [Figure 16] FIG. 10 is a diagram showing median processing of the sensitivity image for the first inspection image. [Figure 17] 10A and 10B are diagrams showing a first example of dead zone processing of the sensitivity image for the first inspection image; [Figure 18] FIG. 10 is a diagram showing a second example of dead zone processing of the sensitivity image for the first inspection image. [Figure 19] 10A and 10B are diagrams illustrating a third example of dead zone processing of the sensitivity image for the first inspection image. [Figure 20] FIG. 10 is a diagram showing a first example of the relationship between the luminance value of the sensitivity image for the first inspection image and the detection threshold value. [Figure 21] FIG. 10 is a diagram showing a second example of the relationship between the luminance value of the sensitivity image for the first inspection image and the detection threshold value. [Figure 22] FIG. 10 is a diagram showing a third example of the relationship between the luminance value of the sensitivity image for the first inspection image and the detection threshold value. [Figure 23] FIG. 10 is a block diagram showing an example of the functional configuration of a processing unit included in an image inspection device according to a second embodiment. [Figure 24] FIG. 10 is a flowchart showing the operation of the image inspection device according to the second embodiment. [Figure 25] FIG. 10 is a diagram showing an example of a processed image for a second inspection image. [Figure 26] FIG. 26 is a diagram showing an example of an inspection image defect candidate area image obtained from the processed image for the second inspection image of FIG. 25. [Figure 27] FIG. 11 is a block diagram showing an example of the functional configuration of a processing unit included in an image inspection device according to a third embodiment. [Figure 28] FIG. 11 is a flowchart of an example of the operation of the image inspection device according to the third embodiment. [Figure 29] FIG. 11 is a diagram showing an example of processing by a processing unit included in an image inspection device according to the third embodiment. [Figure 30] FIG. 10 is a diagram showing a first voting defect area in the image inspection device according to the third embodiment. [Figure 31] FIG. 10 is a diagram showing a second voting defect area in the image inspection device according to the third embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0008] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In each drawing, the same components are designated by the same reference numerals, and duplicated descriptions will be omitted where appropriate.

[0009] The embodiments described below exemplify an image inspection device, an image forming device, and an image inspection method for embodying the technical concept of the present invention, and are not intended to limit the present invention to the embodiments described below. Unless otherwise specified, the shapes of the components, their relative locations, parameter values, etc. described below are intended for illustrative purposes only and are not intended to limit the scope of the present invention. Furthermore, the sizes and positional relationships of components shown in the drawings may be exaggerated for clarity.

[0010] Below, we will explain as an example a case where an image forming device such as a commercial printing machine (production printing machine) that prints a large number of sheets continuously in a short period of time inspects defects in a printed image printed on a recording medium such as paper using an image inspection device related to an embodiment.

[0011] In this specification, image formation and printing are synonymous. A printed image refers to a toner image formed on a recording medium. The recording medium on which the printed image is formed is referred to as an inspection object. An inspection image refers to an image of electronic data captured by an imaging unit, and a non-defective image also refers to an image of electronic data.

[0012] [Embodiment] <Overall Configuration of Image Forming Apparatus 1> Fig. 1 is a diagram showing an example of the overall configuration of an image forming apparatus 1 according to an embodiment. Fig. 1 shows a see-through view of the inside of the image forming apparatus 1.

[0013] The image forming apparatus 1 has an image forming unit 100, an image inspection device 200, and a stacker 300. The image forming unit 100 has an operation panel 101, tandem electrophotographic imaging units 103Y, 103M, 103C, and 103K, a transfer belt 105, a secondary transfer roller 107, a paper feed unit 109, a conveying roller pair 102, a fixing roller 104, and a reverse path 106.

[0014] The operation panel 101 is an operation display unit that inputs various operations to the image forming unit 100 and the image inspection device 200 and displays various screens.

[0015] The image forming units 103Y, 103M, 103C, and 103K each form a toner image through an image forming process (charging process, exposure process, development process, transfer process, and cleaning process) and transfer the formed toner image to the transfer belt 105. In the image forming unit 100, a yellow toner image is formed on the image forming unit 103Y, a magenta toner image is formed on the image forming unit 103M, a cyan toner image is formed on the image forming unit 103C, and a black toner image is formed on the image forming unit 103K. However, this is not limited to this, and the order in which the image forming units 103Y, 103M, 103C, and 103K are arranged may be changed as appropriate. The image forming unit 100 may also have an image forming unit that forms toner images of colors other than yellow, magenta, cyan, and black. Colors other than yellow, magenta, cyan, and black include white, etc.

[0016] The transfer belt 105 transports the full-color toner images transferred and superimposed by the image forming units 103Y, 103M, 103C, and 103K to a secondary transfer position of the secondary transfer roller 107. First, a yellow toner image is transferred (primary transfer) onto the transfer belt 105, followed by a magenta toner image, a cyan toner image, and a black toner image, which are sequentially transferred and superimposed. However, this is not a limitation, and the order in which the toner images of each color are transferred onto the transfer belt 105 may be changed as appropriate. For ease of explanation, hereinafter, unless there is a need to distinguish between colors, the image forming units 103Y, 103M, 103C, and 103K will be collectively referred to as the image forming unit 103.

[0017] Paper feed unit 109 accommodates a plurality of recording media stacked on top of each other and feeds the recording media. Examples of recording media include recording paper (transfer paper), but are not limited to this, and may be coated paper, cardboard, an OHP (Overhead Projector) sheet, plastic film, prepreg, copper foil, or the like, as long as the medium is capable of forming (recording) an image.

[0018] The pair of conveying rollers 102 conveys the recording medium fed from the paper feed unit 109 on a conveying path a in the direction of arrow s. The secondary transfer roller 107 transfers the full-color toner image conveyed by the transfer belt 105 all at once (secondary transfer) at a secondary transfer position onto the recording medium conveyed by the pair of conveying rollers 102. The fixing roller 104 fixes the full-color toner image to the recording medium by applying heat and pressure to the recording medium onto which the full-color toner image has been transferred.

[0019] In the case of single-sided printing, the image forming unit 100 sends the recording medium on which the full-color toner image is fixed to the image inspection device 200. On the other hand, in the case of double-sided printing, the image forming unit 100 sends the recording medium on which the full-color toner image is fixed to the reversing path 106.

[0020] The reverse path 106 switches back the fed recording medium, inverting the front and back sides of the recording medium and transporting it in the direction of arrow t. The recording medium transported by the reverse path 106 is transported again by the transport roller pair 102, and a full-color toner image is transferred onto the opposite side from the previous transfer by the secondary transfer roller 107, and then fixed by the fixing roller 104. The recording medium is then sent to the image inspection device 200 and stacker 300.

[0021] The image inspection device 200 is disposed downstream of the image forming unit 100 in the conveying direction of the recording medium. The image inspection device 200 includes an imaging unit 210, a background unit 220, etc., and inspects defects in a printed image formed on a recording medium (object to be inspected) sent from the image forming unit 100. The image inspection device 200 detects defects in the printed image by comparing the inspection image with a non-defective image. The non-defective image is an image that serves as a sample of the printed image.

[0022] The image forming apparatus 1 processes image data, which is the original data for forming a print image, to create a quality image. For example, information on the printing characteristics of the image forming unit 100 is acquired in advance through experiments or simulations, and the image data is processed by converting the image data using this printing characteristic information. The created quality image is a quality image created through digital processing, and therefore can also be called a digital quality image or a digital master image.

[0023] In addition to the method of processing image data, a sample image can also be read by the imaging unit 210 to create a non-defective image. However, with this method, if the print image is changed frequently, the efficiency of the inspection may decrease because the sample image needs to be read by the imaging unit 210 every time the print image is changed. In contrast, the method of processing image data is advantageous in that it eliminates the need to read the sample image by the imaging unit 210, thereby increasing the efficiency of the inspection.

[0024] Defects in printed images include streaks, dots, stains, scratches, and the like that are visible in printed images. A streak is a linear image area in a printed image that has a different density compared to the surrounding areas. A dot is a small, dot-like image area in a printed image that has a different density compared to the surrounding areas. For example, a dot is a small, dot-like image attached to a white area of ​​the object being inspected. Such streaks, dots, and the like correspond to positive defects that should not be present in a printed image.

[0025] Printed image defects also include blank areas or printing errors in printed images. Blank areas refer to image areas where a toner image should be formed on the object being inspected, but no toner image is formed. Areas where a toner image should be formed correspond to characteristics of a good product, and blank areas correspond to negative defects where the characteristics of a good product are not present on the object being inspected.

[0026] The inspection image is an image obtained by reading and capturing a printed image, which is an object to be inspected, using the imaging unit 210. Details of the configuration of the image inspection device 200 will be described later with reference to FIG.

[0027] The image inspection device 200 discharges the inspection objects, for which inspection of the printed image has been completed, to a stacker 300. The stacker 300 has a tray 301. The stacker 300 stacks the inspection objects discharged from the image inspection device 200 onto the tray 301.

[0028] <Hardware configuration of image forming apparatus according to embodiment> Next, the hardware configuration of the image forming apparatus 1 will be described with reference to Fig. 2. Fig. 2 is a block diagram showing an example of the hardware configuration of the image forming apparatus 1 according to the embodiment.

[0029] As shown in FIG. 2, the image forming apparatus 1 includes a controller 910, a short-range communication circuit 920, an engine control unit 930, an operation panel 940, and a network I / F 950.

[0030] Of these, the controller 910 has a CPU 901 which is the main part of the computer, a system memory (MEM-P) 902, a north bridge (NB) 903, a south bridge (SB) 904, an ASIC (Application Specific Integrated Circuit) 906, a local memory (MEM-C) 907 which is a storage unit, an HDD controller 908, and an HD 909 which is also a storage unit. The NB 903 and the ASIC 906 are connected by an AGP (Accelerated Graphics Port) bus 921.

[0031] Of these, the CPU 901 is a control unit that performs overall control of the image forming apparatus 1. The NB 903 is a bridge that connects the CPU 901 with the MEM-P 902, the SB 904, and the AGP bus 921, and includes a memory controller that controls reading and writing to and from the MEM-P 902, a PCI (Peripheral Component Interconnect) master, and an AGP target.

[0032] The MEM-P 902 comprises a ROM 902a, which is a memory for storing programs and data that realize the functions of the controller 910, and a RAM 902b, which is used as a memory for expanding programs and data and for drawing during memory printing.

[0033] The program stored in RAM 902b may be provided by being recorded on a computer-readable recording medium such as a CD-ROM, CD-R, or DVD in an installable or executable format.

[0034] The SB 904 is a bridge for connecting the NB 903 with PCI devices and peripheral devices. The ASIC 906 is an integrated circuit (IC) for image processing purposes that has hardware elements for image processing, and serves as a bridge for connecting the AGP bus 921, PCI bus 922, HDD 908, and MEM-C 907.

[0035] This ASIC 906 consists of a PCI target and AGP master, an arbiter (ARB) that forms the core of ASIC 906, a memory controller that controls MEM-C 907, multiple DMACs (Direct Memory Access Controllers) that perform image data rotation using hardware logic, etc., and a PCI unit that transfers data between scanner unit 931 and printer unit 932 via PCI bus 922.

[0036] The ASIC 906 may be connected to a USB interface or an IEEE1394 (Institute of Electrical and Electronics Engineers 1394) interface.

[0037] The MEM-C907 is a local memory used as an image buffer for copying and a code buffer. The HD909 is a storage for storing image data, font data used during printing, and forms. The HD909 controls the reading and writing of data from and to the HD909 under the control of the CPU901.

[0038] The AGP bus 921 is a bus interface for a graphics accelerator card proposed to speed up graphics processing, and by directly accessing the MEM-P 902 at high throughput, the graphics accelerator card can be made faster.

[0039] The short-range communication circuit 920 also includes a communication circuit 920a. The communication circuit 920a is a communication circuit such as NFC or Bluetooth (registered trademark).

[0040] Furthermore, the engine control unit 930 has a scanner unit 931 and a printer unit 932. The operation panel 940 has a panel display unit 940a such as a touch panel that displays current setting values, selection screens, etc. and accepts inputs from the operator, and an operation unit 940b that includes a numeric keypad that accepts setting values ​​for image formation conditions such as density setting conditions, a start key that accepts a copy start instruction, etc.

[0041] The controller 910 controls the entire image forming apparatus 1, for example, controlling drawing, communication, and input from the operation panel 940. The scanner unit 931 or the printer unit 932 includes an image processing unit that performs error diffusion, gamma conversion, and the like.

[0042] The image forming apparatus 1 can be used to sequentially switch between a document box function, a copy function, a printer function, and a facsimile function using an application switching key on the operation panel 940.

[0043] When the document box function is selected, the document box mode is selected; when the copy function is selected, the copy mode is selected; when the printer function is selected, the printer mode is selected; and when the facsimile mode is selected, the facsimile mode is selected.

[0044] The network I / F 950 is an interface for performing data communication over a network. The short-range communication circuit 920 and the network I / F 950 are electrically connected to the ASIC 906 via a PCI bus 922.

[0045] Here, the image forming apparatus 1 further includes a lighting unit drive circuit 960, a revolver drive circuit 970, and a pixel array drive circuit 980.

[0046] The lighting unit drive circuit 960 is an electric circuit electrically connected to the lighting unit 240 and drives the lighting unit 240. The lighting unit drive circuit 960 outputs a drive signal to the lighting unit 240 in response to a control signal from the CPU 901 or the like, thereby controlling the intensity and timing of light that the lighting unit 240 irradiates onto the inspection object P.

[0047] The revolver drive circuit 970 is an electric circuit that is electrically connected to the revolver motor 221a attached to the revolver 221 and drives the revolver motor 221a. When the revolver drive circuit 970 outputs a drive signal to the revolver motor 221a in response to a control signal from the CPU 901 or the like, the revolver 221 is driven to rotate, and a predetermined roller from the white small diameter roller 222, the white large diameter roller 224, the black small diameter roller 223, and the black large diameter roller 225 can be brought into contact with the other surface of the object P to be inspected.

[0048] The pixel array drive circuit 980 is an electric circuit that is electrically connected to the pixel array 215 and drives the pixel array 215. An image signal from the pixel array 215 is input via the pixel array drive circuit 980, and can be subjected to predetermined processing or stored in the HD 909 or the like.

[0049] <Configuration of image inspection device 200> The configuration of the image inspection device 200 will be described with reference to Fig. 3. Fig. 3 is a diagram showing an example of the configuration of the image inspection device 200.

[0050] The image inspection device 200 includes an imaging unit 210 that captures an inspection image of an object to be inspected, and a processing unit 260 that processes the inspection image captured by the imaging unit 210. In the example shown in Fig. 3, the image inspection device 200 also includes a contact glass 230, an illumination unit 240, and conveying roller pairs 250 and 251.

[0051] The conveying roller pair 250 and 251 conveys the inspection object P sent from the image forming unit 100 in the Y direction in Fig. 3. One of the conveying roller pair 250 and 251 can be a driving roller pair that is rotationally driven by a driving unit such as a motor, and the other can be a driven roller pair that rotates in accordance with the inspection object P being conveyed.

[0052] The contact glass 230 is made of transparent glass, comes into contact with the transported inspection object P, and has the function of suppressing flapping of the inspection object P when it is read (imaged) by the imaging unit 210. As shown in FIG. 3 , the inspection object P is transported in the Y direction between the contact glass 230 and the background unit 220.

[0053] The lighting unit 240 is configured with an LED array or the like in which multiple LEDs (Light Emitting Diodes) are arranged in the axial direction of the conveying roller pair 250, etc. (the X direction in Figure 3, hereinafter referred to as the width direction), and illuminates a line of light onto the inspection object P being conveyed.

[0054] However, the lighting unit 240 is not limited to this configuration, and may simultaneously light red, green, and blue LEDs and mix the light of each color to irradiate light having a wide wavelength band close to white light. Alternatively, the lighting unit 240 may have a configuration including a single element that irradiates a long linear light in the width direction, like a fluorescent tube. A fluorescent tube can irradiate white light with uniform brightness in the width direction.

[0055] Furthermore, a light-guiding member with its longitudinal direction in the width direction may be used, and white, red, green, and blue LEDs arranged on both ends of the light-guiding member may be turned on to emit light in a line shape by passing the light through the light-guiding member. The light-guiding member allows for the emission of light with uniform brightness in the width direction. A light-guiding lens may also be provided to efficiently guide the light from the LED array to the area through which the edge of the transported inspection object P passes in the width direction.

[0056] The imaging unit 210 is provided on one side (the positive Z direction side in FIG. 3) of the inspection object P that contacts the contact glass 230, and is realized by a CIS (contact image sensor) or the like. More specifically, the imaging unit 210 has mirrors 211-213, a lens 214, and a pixel array 215. Light irradiated from the illumination unit 240 is reflected from the inspection object P, and is reflected by the mirrors 211-213, respectively, as shown by the dashed lines in FIG. 3, and is then imaged by the lens 214 on the light receiving surface of the pixel array 215.

[0057] The pixel array 215 is an element in which PDs (Photo Diodes), which are photoelectric conversion elements that convert optical signals into electrical signals, are arranged in an array in the width direction. One photoelectric conversion element corresponds to one pixel and outputs an electrical signal according to the amount of received light. The pixel array 215 outputs electrical signals (image signals) for one line of pixels in the width direction. At this time, the pixel array 215 also receives reflected light from the inspection object P transported in the Y direction by the transport roller pair 250 and 251 at predetermined timings, and outputs one line of image signals. Two-dimensional image data is acquired by connecting the image signals for one line thus output in a direction perpendicular to the arrangement direction of the pixels in the pixel array 215.

[0058] The pixel array 215 also includes a pixel array 215R that receives red light, a pixel array 215G that receives green light, and a pixel array 215B that receives blue light, each of which is arranged such that the width direction and the pixel arrangement direction are approximately parallel.

[0059] The pixel array 215R that receives red light has a red color filter in front of its light receiving surface and receives red light that passes through the color filter. The red color filter passes light in the red wavelength band and absorbs or reflects light in other wavelength bands. Similarly, the pixel array 215G has a green color filter and receives light in the green wavelength band, and the pixel array 215B has a blue color filter and receives light in the blue wavelength band.

[0060] The pixel array may be a CCD (Charge Coupled Device), a CMOS (Complementary Metal-Oxide-Semiconductor), or the like. Alternatively, the pixel array 215 may be configured using a CCD or CMOS area sensor having a two-dimensional pixel array. Furthermore, in order to increase the light collection efficiency of the pixel array 215, a rod lens array or the like may be provided to guide light reflected by the inspection object P to the pixel array 215.

[0061] The imaging unit 210 receives light reflected from the inspection object P and outputs an image signal including an inspection image formed on the inspection object P.

[0062] The background unit 220 contacts the other surface side (negative Z direction side in FIG. 3) of the inspection object P and has a background member that becomes the background of the inspection object P when the imaging unit 210 reads the edge of the inspection object P. Note that the "other surface" refers to the surface opposite to the surface of the inspection object P on the side where the imaging unit 210 is arranged.

[0063] Background unit 220 has revolver 221, white small diameter roller 222, black small diameter roller 223, white large diameter roller 224, and black large diameter roller 225. White small diameter roller 222, white large diameter roller 224, black small diameter roller 223, and black large diameter roller 225 are attached to revolver 221 so as to be arranged around the cylindrical axis of a cylindrical member included in revolver 221.

[0064] A plurality of circular through-holes that penetrate in the axial direction of the cylinder are formed in the cylindrical member included in revolver 221. Each through-hole is formed so as to be arranged around the cylindrical axis of revolver 221. Each roller can be attached to revolver 221 by passing it through this through-hole.

[0065] The circular through-hole does not necessarily have to have a circular cross-sectional shape, but may have a cross-sectional shape that is a part of a circle. Furthermore, when a prismatic member or the like is attached instead of a roller, a rectangular through-hole may be formed.

[0066] The revolver 221 is rotatable around its cylindrical axis (in the direction of the arrow u in FIG. 3). As an example, by driving a motor attached to the revolver 221 to rotate in the direction of the arrow u using a control signal, the revolver 221 can be rotated in the direction of the arrow u, and a predetermined roller of the plurality of rollers attached to the revolver 221 can be brought into contact with the other surface of the inspection object P. However, the present invention is not limited to rotating the revolver 221 in response to a control signal, and an operator of the image forming apparatus 1 or the like may manually rotate the revolver 221 to bring a predetermined roller of the plurality of attached rollers into contact with the other surface of the inspection object P.

[0067] In this way, the white small diameter roller 222, the white large diameter roller 224, the black small diameter roller 223, and the black large diameter roller 225 are arranged so as to be able to come into contact with the other surface of the inspection object P by the rotation of the revolver 221.

[0068] The white small-diameter roller 222 and the black small-diameter roller 223 have the same diameter but different colors. For example, if the background color of the test object P is white, contacting the black small-diameter roller 223 with the other side of the test object P increases the color contrast between the test object P and the black small-diameter roller 223 serving as a background member, making it easier to detect the edge position of the test object P. While the white small-diameter roller 222 and the black small-diameter roller 223 have been described above as an example, the same applies to the white large-diameter roller 224 and the black large-diameter roller 225. Furthermore, while white and black rollers have been used as an example, this is not limiting and rollers of other colors may be used depending on the color of the test object P.

[0069] On the other hand, the black small-diameter roller 223 and the black large-diameter roller 225 have the same roller color but different roller diameters. Therefore, when the black large-diameter roller 225 is brought into contact with the other side of the inspection object P, the black large-diameter roller 225 pushes the inspection object P in the positive Z direction in Fig. 3, so that the height in the direction intersecting the surface of the inspection object P (the Z direction in Fig. 3) can be made different compared to when the black small-diameter roller 223 is brought into contact with the other side of the inspection object P. In other words, when the black large-diameter roller 225 is brought into contact with the other side of the inspection object P, one side of the inspection object P can be brought closer to the imaging unit 210 compared to when the black small-diameter roller 223 is brought into contact with the other side of the inspection object P.

[0070] The thickness of the inspection object P may vary depending on the type, and the distance (height) from one surface of the inspection object P to the imaging unit 210 differs between a thin inspection object P and a thick inspection object P. Due to such a difference in height, the image read by the imaging unit 210 may contain out-of-focus.

[0071] In particular, the imaging unit 210 has a shallow depth of field because it is configured to be thin. Therefore, the read image is likely to be out of focus due to a slight difference in the height from one side of the inspection object P to the imaging unit 210 caused by a difference in the thickness of the inspection object P. If an out-of-focus image is used, it becomes difficult to accurately detect the edge positions of the inspection object P and the edge positions of the image formed on the inspection object P.

[0072] In contrast to this, in the image inspection device 200, for example, in the case of a thin inspection object P, one side of the inspection object P can be brought close to the imaging unit 210 by bringing the black large-diameter roller 225 into contact with the other side of the inspection object P. Conversely, in the case of a thick inspection object P, one side of the inspection object P can be prevented from coming close to the imaging unit 210 by bringing the black small-diameter roller 223 into contact with the other side of the inspection object P.

[0073] In this way, the height from one side of the inspection object P to the imaging unit 210 can be made constant regardless of the thickness of the inspection object P, and it is possible to prevent out-of-focus in the read image. Although the above has been described using the black small diameter roller 223 and the black large diameter roller 225 as examples, the same applies to the white small diameter roller 222 and the white large diameter roller 224.

[0074] 3, a roller is used as the background member, and the "height" is varied by changing the diameter of the roller, but this is not limiting. For example, the "height" can be varied by using a rectangular pillar as the background member and varying the height (thickness) dimension of the cross-sectional shape of the rectangular pillar.

[0075] The processing unit 260 executes a process of detecting defects in the printed image by comparing an inspection image obtained by capturing a printed image formed on the inspection object P sent from the image forming unit 100 with a non-defective product image. Details of the function of the processing unit 260 will be described below.

[0076] [First embodiment] <Example of functional configuration of processing unit 260> The functional configuration of the processing unit 260 included in the image inspection device 200 according to the first embodiment will be described with reference to Fig. 4. Fig. 4 is a block diagram showing an example of the functional configuration of the processing unit 260.

[0077] Processing unit 260 includes inspection image segmented image generator 261, which generates inspection image segmented images using a first inspection image processed image obtained by averaging pixel brightness values ​​in each of a plurality of processing regions obtained by dividing the inspection image according to predetermined inspection image division conditions. Processing unit 260 also includes a good-quality image segmented image generator 262, which generates good-quality image segmented images using a first good-quality image processed image obtained by averaging pixel brightness values ​​in each of a plurality of processing regions obtained by dividing a good-quality image, to be compared with the inspection image, according to predetermined inspection image division conditions. Processing unit 260 also includes a sensitivity map image generator 263, which generates a sensitivity map image of the same size as the inspection image and the good-quality image, in which a detection threshold is set for each predetermined detection region. Processing unit 260 also includes a first inspection image sensitivity image generator 264, which divides the sensitivity map image into a plurality of regions according to the first inspection image division conditions, and generates a first inspection image sensitivity image by processing the first inspection image threshold of each region according to the detection threshold. The processing unit 260 also includes a first-good-image sensitivity image generating unit 265 that divides the sensitivity map image into multiple regions according to the first-good-image division condition and generates a first-good-image sensitivity image by processing each region using the first-good-image threshold value in accordance with the detection threshold. The processing unit 260 also includes an inspection-image defect candidate extracting unit 266 that compares the luminance values ​​of pixels in a region of interest for the inspection image among the multiple processing regions included in the divided image for the inspection image with the luminance values ​​of pixels in processing regions surrounding the region of interest for the inspection image, and extracts a defect candidate region for the inspection image based on the first inspection-image threshold. The processing unit 260 also includes a good-image defect candidate extracting unit 267 that compares the luminance values ​​of pixels in a region of interest for the good image among the multiple processing regions included in the divided image for the good image with the luminance values ​​of pixels in processing regions surrounding the region of interest for the good image, and extracts a defect candidate region for the good image based on the first good-image threshold. The processing unit 260 also has a defect detection unit 268 that compares defect candidate areas for the inspection image with defect candidate areas for the non-defective image to detect defect areas, and an output unit 269 that outputs the detection results of the defect areas by the defect detection unit 268.

[0078] The image inspection device 200 extracts defect candidates using a segmented image for an inspection image and a segmented image for a non-defective product image, which have been reduced in resolution by averaging the brightness values ​​of each region. The image inspection device 200 models the mechanism by which humans unconsciously recognize differences from a normal state, and can extract defect candidates in the inspection object P at high speed and with high sensitivity. The image inspection device 200 can detect positive and negative defects at high speed and with high sensitivity by comparing defect candidates in the inspection image with defect candidates in the non-defective product image.

[0079] In a typical image inspection device, for example, if the same detection threshold is used for defect detection in an inspection image and a non-defective image between an area where pixel values ​​change significantly, such as a pattern or edge, and an area where pixel values ​​change only slightly, such as the background, the defect detection accuracy may be reduced. Specifically, if a low-sensitivity detection threshold is used to suppress false detections in an area where pixel values ​​change significantly, sufficient detection accuracy cannot be achieved in an area where pixel values ​​change only slightly. On the other hand, if a high-sensitivity detection threshold is used to achieve sufficient defect detection accuracy in an area where pixel values ​​change only slightly, the number of false detections increases in an area where pixel values ​​change only slightly.

[0080] In response to the above, it is conceivable to suppress a decrease in defect detection accuracy by setting different detection thresholds for regions with large changes in pixel values ​​and regions with small changes in pixel values. For example, by setting a low-sensitivity detection threshold for regions with large changes in pixel values ​​and a high-sensitivity detection threshold for regions with small changes in pixel values, it is possible to suppress false detection in regions with large changes in pixel values ​​and obtain sufficient defect detection accuracy in regions with small changes in pixel values.

[0081] However, when detecting defects using the reduced-resolution inspection image segmented images and non-defective image segmented images, the detection thresholds for each region are also set at a lower resolution. This may result in a decrease in the accuracy of defect detection. For example, in regions where pixel values ​​change significantly, the effect of suppressing false detection may be reduced, or in regions where pixel values ​​change only slightly, sufficient defect detection accuracy may not be achieved.

[0082] The image inspection device 200 according to the embodiment generates a sensitivity map image in which a detection threshold is set for each detection region using the sensitivity map image generation unit 263. The image inspection device 200 also reduces the resolution of the sensitivity map image using the first inspection image sensitivity image generation unit 264, in the same way as for the inspection image, according to the division conditions for the first inspection image. The image inspection device 200 also reduces the resolution of the sensitivity map image using the first non-defective image sensitivity image generation unit 265, in the same way as for the non-defective image, according to the division conditions for the first non-defective image.

[0083] The image inspection device 200 extracts defect candidate areas for the inspection image using a sensitivity map image for the inspection image with reduced resolution, and extracts defect candidate areas for the good product image using a sensitivity map image for the good product image with reduced resolution. The image inspection device 200 detects defect areas using the defect candidate areas for the inspection image and the defect candidate areas for the good product image. As described above, the image inspection device 200 can suppress a decrease in defect detection accuracy that accompanies a reduction in the resolution of the sensitivity map image by matching the resolution of the divided images for the inspection image and the divided images for the good product image with the resolution of the sensitivity map image. For example, the image inspection device 200 can suppress false detection in areas where pixel values ​​change significantly, and achieve sufficient defect detection accuracy in areas where pixel values ​​change only slightly.

[0084] As described above, this embodiment can provide an image inspection device 200 and an image inspection method that can inspect an inspection object P at high speed and with high accuracy. It can also provide an image forming device 1 that has an image inspection device 200 that can inspect an inspection object P at high speed and with high accuracy. The functional configuration of the processing unit 260 will be described in detail below.

[0085] Each functional configuration shown in Fig. 4 is a function or a means for performing a function that is realized when any of the components shown in Fig. 2 operates in response to an instruction from CPU 901 in accordance with a program loaded from ROM 902a onto RAM 902b. Note that Fig. 4 shows the main components of processing unit 260, but processing unit 260 may have other components.

[0086] The predetermined division conditions for the first inspection image in the divided image generation unit 261 for inspection images are division conditions such as the number of divisions, the size or shape of the divided regions, etc. "Each of the multiple processing regions into which the inspection image is divided" refers to the region corresponding to each lattice when the inspection image is divided into multiple regions using a matrix lattice. The processed image for the first inspection image is a mosaic-like low-resolution image consisting of multiple regions in which the luminance values ​​of all pixels included in one region are replaced with the average luminance value of all pixels included in that region.

[0087] The predetermined division conditions for the first good product image in the divided image generation unit for good product image 262 are division conditions such as the number of divisions, the size or shape of the divided regions, etc. The division conditions for the first good product image may be the same as or different from the division conditions for the first inspection image. The processed image for the first good product image is a mosaic-like low-resolution image made up of multiple regions in which the luminance values ​​of all pixels included in one region are replaced with the average luminance value of all pixels included in one region.

[0088] In the example shown in FIG. 4, a good-quality image is created in advance and stored in HD 909 or the like shown in FIG. 2. The good-quality image divided image generation unit 262 acquires the good-quality image by referring to HD 909 or the like. A good-quality image is, for example, an image of a good-quality inspection object. By using an image of a good-quality inspection object as the good-quality image, the good-quality image can be easily acquired. However, the good-quality image is not limited to an image of a good-quality inspection object. For example, the good-quality image may be a digital master image generated based on image data that is the source of the inspection image. Using the digital master image as the good-quality image makes it possible to change and adjust image characteristics such as color and resolution, increasing the freedom of selection of good-quality images.

[0089] The detection threshold in sensitivity map image generation unit 263 is set based on, for example, the luminance value of the inspection image, the luminance difference between pixels in a region of interest for the inspection image and pixels in processing regions surrounding the region of interest for the inspection image among multiple processing regions included in the divided image for the inspection image, or an input by the user of image inspection device 200. By setting the detection threshold in this manner, an appropriate detection threshold is set according to the change in pixel value of the region. For example, it is possible to set a low-sensitivity detection threshold in a region where the change in pixel value is large, and a high-sensitivity detection threshold in a region where the change in pixel value is small.

[0090] The detection threshold in the sensitivity image generation unit 264 for the first examination image is the detection threshold set by the sensitivity map image generation unit 263. The "processing according to the detection threshold" is an averaging process, a maximization process, a minimization process, a median process, etc.

[0091] The averaging process is a process of setting, for each of a plurality of regions obtained by dividing a sensitivity map image, a detection threshold obtained by averaging a plurality of detection thresholds within the region as a unified detection threshold within the region. The maximization process is a process of setting, for each of a plurality of regions obtained by dividing a sensitivity map image, a detection threshold that is the maximum value among the plurality of detection thresholds within the region as a unified detection threshold within the region. The minimization process is a process of setting, for each of a plurality of regions obtained by dividing a sensitivity map image, a detection threshold that is the minimum value among the plurality of detection thresholds within the region as a unified detection threshold within the region. The medianization process is a process of setting, for each of a plurality of regions obtained by dividing a sensitivity map image, a detection threshold that is the median value among the plurality of detection thresholds within the region as a unified detection threshold within the region.

[0092] The detection threshold in the first non-defective image sensitivity image generating unit 265 is the detection threshold set by the sensitivity map image generating unit 263, similar to the detection threshold in the first inspection image sensitivity image generating unit 264. The "processing according to the detection threshold" is an averaging process, a maximization process, a minimization process, a median process, etc.

[0093] For example, the inspection image defect candidate extraction unit 266 calculates the difference in pixel brightness values ​​between the inspection image region of interest and the region surrounding the inspection image region of interest, and extracts the inspection image region of interest as a defect candidate region if the difference value is equal to or greater than a predetermined threshold. The inspection image defect candidate extraction unit 266 performs the above process on each of the multiple regions in the inspection image divided image while changing the inspection image region of interest, thereby extracting defect candidate regions in the inspection image as inspection image defect candidate regions. The inspection image defect candidate extraction unit 266 can visualize and display the extracted inspection image defect candidate regions by designating the inspection image defect candidate regions as valid regions and the regions other than the inspection image defect candidate regions as invalid regions in the inspection image divided image.

[0094] The non-defective image defect candidate extraction unit 267, for example, calculates the difference in pixel brightness values ​​between the non-defective image region of interest and the region surrounding the non-defective image region of interest, and extracts the non-defective image region of interest as a defect candidate region if the difference value is equal to or greater than a predetermined threshold. The non-defective image defect candidate extraction unit 267 performs the above process on each of the multiple regions in the non-defective image segmented image while changing the non-defective image region of interest, thereby extracting defect candidate regions in the non-defective image as non-defective image defect candidate regions. The non-defective image defect candidate extraction unit 267 can visualize and display the extracted non-defective image defect candidate regions by designating the non-defective image defect candidate regions as valid regions and the regions other than the non-defective image defect candidate regions in the non-defective image segmented image as invalid regions.

[0095] For example, the defect detection unit 268 removes (non-extracts) defect candidate areas that are common to both the inspection image defect candidate areas and the good product image defect candidate areas as characteristics of a good product. The defect detection unit 268 can detect areas extracted only using the inspection image defect candidate areas as positive defect areas that are present in the inspection image but not in the good product image. Furthermore, the defect detection unit 268 can detect areas extracted only using the good product image defect candidate areas as negative defect areas that are present in the good product image but not in the inspection image.

[0096] The output unit 269 outputs the defect detection result in the printed image of the inspection object P as the detection result to a device or equipment other than the processing unit 260. The device or equipment other than the processing unit 260 is an information processing device, a display device, a storage device, a communication device, etc. other than the processing unit 260. The information processing device is a PC (Personal Computer), etc.

[0097] <Example of operation of image inspection device 200> The operation of the image inspection device 200 will be described with reference to FIG. 5. FIG. 5 is a flowchart showing an example of the operation of the image inspection device 200. FIG. 5 shows the operation of the image inspection device 200, which starts when an inspection object P on which a print image has been formed is sent from the image forming unit 100 to the image inspection device 200. However, the operation start condition of the image inspection device 200 is not limited to when the inspection object P is sent from the image forming unit 100 to the image inspection device 200. For example, the image inspection device 200 may start the operation of FIG. 5, which starts when an inspection start operation input is made by a user of the image inspection device 200 or the image forming device 1 using the operation panel 101.

[0098] First, in step S11, the image inspection device 200 uses the imaging unit 210 to capture an image of a print image formed on the inspection object P as an inspection image.

[0099] Next, in step S12, image inspection device 200 inputs an image signal including the inspection image captured by imaging unit 210 to processing unit 260. Inspection image divided image generation unit 261 of processing unit 260 divides the inspection image into a plurality of regions that are a matrix lattice according to the first inspection image division condition.

[0100] Next, in step S13, image inspection device 200 generates a first processed image for inspection image by using divided image for inspection image generation unit 261, which averages the luminance values ​​of pixels contained in each of the divided regions.

[0101] Next, in step S14, image inspection device 200 generates an inspection image segmented image using the first inspection image processed image by inspection image segmented image generation unit 261. Inspection image segmented image generation unit 261 passes the generated inspection image segmented image to inspection image defect candidate extraction unit 266.

[0102] Subsequently, in step S15, the image inspection device 200 causes the non-defective image divided image generating unit 262 to read and acquire the non-defective image stored in the HD 909 or the like.

[0103] Next, in step S16, the image inspection device 200 causes the divided image generation unit 262 for a non-defective image to divide the non-defective image into a plurality of regions that are a matrix lattice in accordance with the first division condition for a non-defective image.

[0104] Next, in step S17, the image inspection device 200 generates a first processed image for a good product image by using the divided image generation unit 262 for a good product image, which averages the brightness values ​​of the pixels contained in each of the divided regions.

[0105] Next, in step S18, the image inspection device 200 generates a segmented image for a good product image using the first processed image for a good product image by the segmented image generation unit for a good product image 262. The segmented image generation unit for a good product image 262 passes the generated segmented image for a good product image to the defect candidate extraction unit for a good product image 267.

[0106] Next, in step S19, the image inspection device 200 causes the sensitivity map image generating unit 263 to generate a sensitivity map image.

[0107] Next, in step S20, the image inspection device 200 divides the sensitivity map image into multiple regions according to the division conditions for the first inspection image using the first inspection image sensitivity image generation unit 264. The first inspection image sensitivity image generation unit 264 passes the first inspection image sensitivity image to the inspection image defect candidate extraction unit 266.

[0108] Next, in step S21, the image inspection device 200 divides the sensitivity map image into a plurality of regions according to the division conditions for the first good product image by the first good product image sensitivity image generation unit 265, and generates a first good product image sensitivity image. The first good product image sensitivity image generation unit 265 passes the first good product image sensitivity image to the good product image defect candidate extraction unit 267.

[0109] Next, in step S22, image inspection device 200 compares the brightness values ​​of pixels in the region of interest for inspection image among the multiple processing regions included in the divided image for inspection image with the brightness values ​​of pixels in processing regions surrounding the region of interest for inspection image, and extracts defect candidate regions for inspection image based on the first threshold value for inspection image. Inspection image defect candidate extraction unit 266 passes the extracted defect candidate regions for inspection image to defect detection unit 268.

[0110] Next, in step S23, the image inspection device 200 compares the brightness values ​​of pixels in the region of interest for a good product image, among the multiple processing regions included in the divided image for a good product image, with the brightness values ​​of pixels in processing regions around the region of interest for a good product image, and extracts a defect candidate region for a good product image based on the first threshold value for a good product image. The defect candidate extraction unit 267 for a good product image passes the extracted defect candidate region for a good product image to the defect detection unit 268.

[0111] Next, in step S24, the image inspection device 200 compares the defect candidate areas for the inspection image with the defect candidate areas for the non-defective image to detect a defect area using the defect detection unit 268. The defect detection unit 268 passes the defect area detection result to the output unit 269.

[0112] Subsequently, in step S25, the image inspection device 200 outputs the detection result of the defective area to a device or equipment other than the processing unit 260 via the output unit 269.

[0113] In this manner, the image inspection device 200 can inspect the inspection object P for defects.

[0114] In FIG. 5, the operations on the non-defective image from step S15 to step S18 may be performed before the operations on the inspection image from step S11 to step S14, and the operations on the non-defective image and the operations on the inspection image may be performed in parallel.

[0115] 5 may be performed at any timing from step S11 to step S18. The operation of step S21 may be performed before the operation of step S20, or the operations of step S21 and step S20 may be performed in parallel. The operation of step S23 may be performed before the operation of step S22, or the operations of step S23 and step S22 may be performed in parallel.

[0116] <Processing Results by Processing Unit 260> FIG. 6 is a diagram showing an example of an inspection image 60. FIG. 7 is a diagram showing an example of a first processed image for inspection image 61. FIG. 8 is a diagram showing an example of an inspection image defect candidate area image 63 obtained from the first processed image for inspection image 61 of FIG. 7. FIG. 9 is a diagram showing an example of a non-defective image 160. FIG. 10 is a diagram showing an example of a first processed image for non-defective image 161. FIG. 11 is a diagram showing an example of a non-defective image defect candidate area image 163 obtained from the first processed image for non-defective image 161 of FIG. 10.

[0117] Inspection image 60 shown in Figure 6 includes defect 70 and non-defective feature 80. First processed image for inspection image 61 shown in Figure 7 is a mosaic image obtained by averaging the pixel brightness values ​​in each of the multiple processing regions into which inspection image 60 is divided. First processed image for inspection image 61 includes defect region 71 corresponding to defect 70 and non-defective feature region 81 corresponding to non-defective feature 80. A segmented image for inspection image is obtained using first processed image for inspection image 61.

[0118] In inspection image defect candidate area image 63 shown in Fig. 8, areas other than the inspection image defect candidate area are displayed in black as invalid areas. Inspection image defect candidate area image 63 shown in Fig. 8 includes inspection image defect candidate area 73 corresponding to defect 70 and inspection image defect candidate area 83 corresponding to non-defective product feature 80.

[0119] 9 includes a defect 170 and a non-defective feature 180. The defect 170 is not included in the inspection image 60, and corresponds to a negative defect in which the inspection object P does not have a feature of a non-defective product that should be present.

[0120] 10 is a mosaic-like image obtained by averaging the pixel brightness values ​​in each of a plurality of processing regions into which the good product image 160 is divided. The first processed image for good product image 161 includes a defect region 171 corresponding to the defect 170 and a good product feature region 181 corresponding to the good product feature 180. A segmented image for good product image is obtained using the first processed image for good product image 161.

[0121] In the defect candidate area image for a non-defective product image 163 shown in Fig. 11, areas other than the defect candidate areas for a non-defective product image are displayed in black as invalid areas. The defect candidate area image for a non-defective product image 163 shown in Fig. 11 includes a defect candidate area for a non-defective product image 173 corresponding to the defect 170 and a defect candidate area for a non-defective product image 183 corresponding to the non-defective product feature 180.

[0122] The defect detection unit 268 compares, for example, the defect candidate area image for inspection image 63 of FIG. 8 with the defect candidate area image for non-defective image 163 of FIG. 11, and removes (non-extracts) defect candidate areas common to both, assuming that they are characteristics of a non-defective product. The defect detection unit 268 detects areas extracted only in the defect candidate area image for inspection image 63 as positive defect areas that are present in the inspection image 60 but not in the non-defective product image 160. The defect detection unit 268 also detects areas extracted only in the defect candidate area image for non-defective product image 163 as negative defect areas that are present in the non-defective product image 160 but not in the inspection image 60. In this way, the image inspection device 200 can detect both positive and negative defects.

[0123] 12 is a diagram illustrating the sensitivity map image 75 and the sensitivity image for the first inspection image 85. FIG. 12 shows the inspection image 60, the sensitivity map image 75, and the sensitivity image for the first inspection image 85.

[0124] Sensitivity map image 75 is an image of the same size as inspection image 60, in which a detection threshold is set for each predetermined detection region. Sensitivity image 85 for first inspection image is an image obtained by dividing sensitivity map image 75 into multiple regions according to the division conditions for the first inspection image, and processing the threshold for the first inspection image for each region according to the detection threshold.

[0125] The sensitivity image for the first inspection image 85-1, the sensitivity image for the first inspection image 85-2, and the sensitivity image for the first inspection image 85-3 are sensitivity images for the first inspection image with different resolutions. The sensitivity image for the first inspection image 85-1 has the highest resolution, followed by the sensitivity image for the first inspection image 85-2 and the sensitivity image for the first inspection image 85-3. Note that the sensitivity image for the first inspection image 85 is a general term used when no distinction is made between the sensitivity image for the first inspection image 85-1, the sensitivity image for the first inspection image 85-2, and the sensitivity image for the first inspection image 85-3.

[0126] The sensitivity map image 75 and the sensitivity image 85 for the first inspection image each include a dead zone region 76, a low sensitivity region 77, and a high sensitivity region 78. The dead zone region 76 is a region that is not subject to defect detection and is a region where no detection threshold is set. The low sensitivity region 77 and the high sensitivity region 78 are regions that are subject to defect detection. The low sensitivity region 77 is a region where pixel values ​​change greatly, including patterns and edges, and is a region where a high detection threshold is set. The high sensitivity region 78 is a region where pixel values ​​change little, including backgrounds, and is a region where a low detection threshold is set.

[0127] The first inspection image sensitivity image 85 is created by dividing the inspection image 60 according to the first inspection image division conditions, and then dividing the sensitivity map image 75 according to the first inspection image division conditions to match the resolution-reduced inspection image divided images. This allows the resolution of the first inspection image sensitivity image 85 to match the resolution of the inspection image divided images. By matching the resolution of the first inspection image sensitivity image 85 to the resolution of the inspection image divided images, a decrease in defect detection accuracy due to the reduction in resolution of the sensitivity map image 75 is suppressed. Note that while FIG. 12 shows an example corresponding to an inspection image, similar processing can also be applied to images of non-defective products.

[0128] 13 to 19 are diagrams illustrating processing of the sensitivity image 85 for the first inspection image. FIG. 13 is a diagram illustrating averaging processing of the sensitivity image 85 for the first inspection image. FIG. 14 is a diagram illustrating maximum value processing of the sensitivity image 85 for the first inspection image. FIG. 15 is a diagram illustrating minimum value processing of the sensitivity image 85 for the first inspection image. FIG. 16 is a diagram illustrating median value processing of the sensitivity image 85 for the first inspection image. FIG. 17 is a diagram illustrating a first example of dead zone processing of the sensitivity image 85 for the first inspection image. FIG. 18 is a diagram illustrating a second example of dead zone processing of the sensitivity image 85 for the first inspection image. FIG. 19 is a diagram illustrating a third example of dead zone processing of the sensitivity image 85 for the first inspection image.

[0129] 13 to 16 each show how the luminance values ​​of four pixels constituting the sensitivity image 85 for the first inspection image are processed. The luminance values ​​of the four pixels are 50, 120, 100, and 200. In the averaging process shown in FIG. 13, the luminance values ​​of the four pixels are averaged to calculate a luminance value of "118," and the luminance values ​​of each of the four pixels are replaced with 118. In the maximum value process shown in FIG. 14, the maximum luminance value of "200" is selected from the four pixels, and the luminance values ​​of each of the four pixels are replaced with 200. In the minimum value process shown in FIG. 15, the minimum luminance value of "50" is selected from the four pixels, and the luminance values ​​of each of the four pixels are replaced with 50. In the median value process shown in FIG. 16, the median luminance value of "110" is selected from the four pixels, and the luminance values ​​of each of the four pixels are replaced with 110.

[0130] 17 to 19 each show how the luminance values ​​of four pixels constituting the sensitivity image 85 for the first inspection image are processed. In the first example shown in FIG. 17, the luminance values ​​of the four pixels are 0, 100, 100, and 200. Pixels with a luminance value of 0 correspond to the dead zone. In the first example, the luminance values ​​of the four pixels are replaced with 0, and the pixels are set to the dead zone. In the second example shown in FIG. 18, the luminance values ​​of the four pixels are 0, 120, 100, and 200. In the second example, the luminance values ​​of the pixels other than the dead zone are averaged to calculate a luminance value of "140," and the luminance values ​​of the four pixels are replaced with 140.

[0131] The third example shown in FIG. 19 shows cases where the brightness values ​​of four pixels are 0, 120, 100, and 200, and cases where they are 0, 0, 0, and 200. When the brightness values ​​of the four pixels are 0, 0, 0, and 200, the dead zone is the majority. When the brightness values ​​of the four pixels are 0, 120, 100, and 200, the dead zone is not the majority. When the dead zone is the majority, all brightness values ​​of the four pixels are set to 0, and the pixels are set to the dead zone. On the other hand, when the dead zone is not the majority, the brightness values ​​of the pixels that are not in the dead zone are averaged, and the brightness values ​​of all four pixels are replaced with the average brightness value of "140."

[0132] Although an example of averaging processing is shown in FIGS. 13 to 19, maximum value processing, minimum value processing, median value processing, etc. may be used instead of averaging processing.

[0133] Fig. 20 is a diagram showing a first example of the relationship between the luminance value of the sensitivity image 85 for the first inspection image and the detection threshold. Fig. 21 is a diagram showing a second example of the relationship between the luminance value of the sensitivity image 85 for the first inspection image and the detection threshold. Fig. 22 is a diagram showing a third example of the relationship between the luminance value of the sensitivity image 85 for the first inspection image and the detection threshold.

[0134] 20, the luminance value of the sensitivity image 85 for the first inspection image and the detection threshold are in a proportional relationship, and the range of values ​​is limited. However, the luminance value of the sensitivity image 85 for the first inspection image and the detection threshold may be in a relationship other than a proportional relationship. The relationship other than a proportional relationship may be expressed by a logarithm, an exponent, a polynomial, or the like.

[0135] The second example shown in FIG. 21 and the third example shown in FIG. 22 show cases where the relationship between the luminance value of the sensitivity image 85 for the first inspection image and the detection threshold is changed depending on the resolution (i.e., the grid size composed of pixels). In the second example shown in FIG. 21, the luminance value of the sensitivity image 85 for the first inspection image is proportional to the detection threshold. The proportionality coefficient differs depending on the grid size. In the third example shown in FIG. 22, for grid sizes def and 2×2, the luminance value of the sensitivity image 85 for the first inspection image is proportional to the detection threshold, but the slope differs. For grid size 3×3, the detection threshold is constant when the luminance value of the sensitivity image 85 for the first inspection image is less than 118, and is proportional to the detection threshold when the luminance value of the sensitivity image 85 for the first inspection image is 119 or greater.

[0136] As shown in Figures 20 to 22, the relationship between the luminance value of the sensitivity image 85 for the first inspection image and the detection threshold can be determined by a mathematical formula. That is, the sensitivity image generation unit 264 for the first inspection image shown in Figure 4 can calculate the threshold for the first inspection image by a mathematical formula based on the detection threshold. By having the sensitivity image generation unit 264 for the first inspection image calculate the threshold for the first inspection image by using a mathematical formula, processing can be simplified compared to using a table that associates the detection threshold with the threshold for the first inspection image. Furthermore, since the storage capacity of the table is not required, the storage capacity of the image inspection device 200 can be reduced.

[0137] 20 to 22 show the processing by the sensitivity image generating unit 264 for the first inspection image, but the sensitivity image generating unit 265 for the first inspection image shown in FIG. 4 can also execute processing similar to the processing by the sensitivity image generating unit 264 for the first inspection image shown in FIGS. 20 to 22. That is, the sensitivity image generating unit 265 for the first inspection image can calculate the threshold value for the first inspection image using a mathematical formula based on the detection threshold value. By calculating the threshold value for the first inspection image using a mathematical formula, the sensitivity image generating unit 265 for the first inspection image can simplify processing compared to using a table or the like that associates the detection threshold value with the threshold value for the first inspection image. Furthermore, since the storage capacity of the table is not required, the storage capacity of the image inspection device 200 can be reduced.

[0138] [Second embodiment] Next, an image forming apparatus according to a second embodiment will be described. Note that the same names and symbols as those in the previously described embodiments indicate the same or similar components or configurations, and detailed descriptions will be omitted as appropriate. This also applies to the following embodiments.

[0139] <Functional configuration of processing unit included in image inspection device according to second embodiment> 23 is a block diagram showing an example of the functional configuration of a processing unit 260a included in an image inspection device 200a according to the second embodiment. The processing unit 260a differs from the processing unit 260 in the first embodiment in that it includes a divided image generation unit for inspection image 261a, a divided image generation unit for non-defective image 262a, a sensitivity image generation unit for second inspection image 270, and a sensitivity image generation unit for second non-defective image 271.

[0140] The inspection image divided image generation unit 261a generates a divided image for the inspection image using at least one of a first inspection image processed image obtained by averaging the pixel brightness values ​​in each of multiple processing areas into which the inspection image is divided according to a predetermined first inspection image division condition, and a second inspection image processed image obtained by averaging the pixel brightness values ​​in each of multiple processing areas into which the inspection image is divided according to a second inspection image division condition that differs from the first inspection image division condition in at least one of phase, direction, and size.

[0141] For example, the inspection image divided image generating unit 261a generates the inspection image divided image by selecting either the first inspection image processed image or the second inspection image processed image. From the viewpoint that the first inspection image processed image becomes the inspection image divided image, and the second inspection image processed image becomes the inspection image divided image, the inspection image divided image can be said to be a general term for the first inspection image processed image and the second inspection image processed image.

[0142] It is preferable that the phase, direction, or size of the areas (grids) applied when generating the original divided image for the inspection image and the divided image for the good product image used for comparison by the defect detection unit 268 be the same for the defect candidate areas for the inspection image and the defect candidate areas for the good product image.

[0143] In the inspection image divided image generation unit 261a, the phase of a region refers to the position of a grid that divides an image, such as an inspection image, into multiple regions. The direction of a region refers to the direction in which the phase of the region is shifted. The direction of a region is basically vertical (the column direction of the grid) or horizontal (the row direction of the grid), but may also include a diagonal direction defined by an angle rotated around the center point of the grid. The size of a region refers to the area of ​​the region. For example, the area of ​​the region is expressed in terms of the number of pixels, such as 10 x 10 pixels. In each of the multiple processing regions obtained by dividing the inspection image according to the second inspection image division conditions, the luminance values ​​of all pixels included in one processing region are replaced with the average luminance value of all pixels included in one processing region. The second inspection image processed image is a mosaic-like low-resolution image consisting of multiple regions.

[0144] The good-quality image divided image generating unit 262a generates a good-quality image divided image using at least one of a first good-quality image processed image obtained by averaging the pixel brightness values ​​in each of a plurality of processing areas obtained by dividing a good-quality image, which is to be compared with the inspection image, according to a predetermined first good-quality image division condition, and a second good-quality image processed image obtained by averaging the pixel brightness values ​​in each of a plurality of processing areas obtained by dividing a good-quality image according to a second good-quality image division condition that differs from the first good-quality image division condition in at least one of phase, direction, or size.

[0145] For example, the good-quality image divided image generating unit 262a generates a good-quality image divided image by selecting either the first good-quality image processed image or the second good-quality image processed image. From the viewpoint that the first good-quality image processed image becomes the good-quality image divided image and the second good-quality image processed image becomes the good-quality image divided image, the good-quality image divided image can be said to be a collective term for the first good-quality image processed image and the second good-quality image processed image.

[0146] In the divided image generating unit for good quality image 262a, in each of a plurality of processing regions obtained by dividing the good quality image according to the second dividing condition for good quality image, the luminance values ​​of all pixels included in one processing region are replaced with the average value of the luminance values ​​of all pixels included in one processing region. The second processed image for good quality image is a mosaic-like low-resolution image made up of such a plurality of regions.

[0147] The image inspection device 200a generates a segmented image for the inspection image using at least one of the first processed image for the inspection image and the second processed image for the inspection image. The image inspection device 200a also generates a segmented image for the good product image using at least one of the first processed image for the good product image and the second processed image for the good product image. By detecting defective areas in the inspection image using the segmented image for the inspection image and the segmented image for the good product image thus generated, the image inspection device 200a can utilize the mechanism by which humans unconsciously recognize differences from normal states and detect the defective areas quickly and with high sensitivity.

[0148] The first inspection image sensitivity image generation unit 264 may calculate the first inspection image threshold value using a mathematical formula based on the detection threshold value, the first non-defective image sensitivity image generation unit 265 may calculate the first non-defective image threshold value using a mathematical formula based on the detection threshold value, the second inspection image sensitivity image generation unit 270 may calculate the second inspection image threshold value using a mathematical formula based on the detection threshold value, and the second non-defective image sensitivity image generation unit 271 may calculate the second non-defective image threshold value using a mathematical formula based on the detection threshold value.

[0149] By calculating the threshold values ​​for the first inspection image, the first non-defective image, the second inspection image, and the second non-defective image using mathematical expressions, processing can be simplified compared to using a table that associates the threshold values ​​for the first inspection image, the first non-defective image, the second inspection image, and the second non-defective image with the detection threshold. Furthermore, since the storage capacity of the table is not required, the storage capacity of the image inspection device 200a can be reduced.

[0150] <Example of operation of image inspection device 200a> The operation of the image inspection device 200a will be described with reference to FIG. 24. FIG. 24 is a flowchart showing an example of the operation of the image inspection device 200a. FIG. 24 shows the operation of the image inspection device 200a, which starts when the inspection object P on which a print image has been formed is sent from the image forming unit 100 to the image inspection device 200a. However, the operation start condition of the image inspection device 200a is not limited to the inspection object P being sent from the image forming unit 100 to the image inspection device 200a. For example, the image inspection device 200a may start the operation of FIG. 24 when the user of the image inspection device 200a or the image forming device 1 inputs an operation to start inspection using the operation panel 101. The operation of the image inspection device 200a will be described, focusing on the differences from the operation of the image inspection device 200 shown in FIG. 5.

[0151] The processing from step S31 to step S33 is the same as the processing from step S11 to step S13 in FIG.

[0152] In step S34, image inspection device 200a causes divided image for inspection image generation unit 261a to change at least one of the phase, direction, and size of each of the divided processing regions relative to the first processed image for inspection image.

[0153] Next, in step S35, the image inspection device 200a divides the inspection image divided image generation unit 261a into multiple processing areas in which at least one of the phase, direction, or size of each processing area is changed relative to the processed image for the first inspection image.

[0154] Next, in step S36, the image inspection device 200a generates a processed image for the second inspection image by using the divided image generation unit 261a for the inspection image to average the brightness values ​​of the pixels contained in each area in the multiple processing areas obtained by division.

[0155] Next, in step S37, image inspection device 200a generates a divided image for inspection image using at least one of the first processed image for inspection image and the second processed image for inspection image, using divided image for inspection image generation unit 261a.

[0156] The processing from step S38 to step S40 is the same as the processing from step S15 to step S17 in FIG.

[0157] In step S41, the image inspection device 200a changes at least one of the phase, direction, and size of each of the divided processing regions in the plurality of divided processing regions with respect to the first processed image for a non-defective image using the divided image for a non-defective image generating unit 262a.

[0158] Next, in step S42, the image inspection device 200a divides the processed image for the first good product image into multiple processing areas by using the divided image generation unit 262a for the good product image, changing at least one of the phase, direction, or size of each processing area relative to the processed image for the first good product image.

[0159] Next, in step S43, the image inspection device 200a generates a second processed image for a good product image by using the divided image generation unit 262a to average the brightness values ​​of the pixels contained in each area in the multiple processing areas obtained by division.

[0160] Subsequently, in step S44, the image inspection device 200a generates a non-defective divided image using at least one of the first non-defective processed image and the second non-defective processed image by the non-defective divided image generating unit 262a.

[0161] The processing from step S45 to step S51 is the same as the processing from step S19 to step S25.

[0162] In this manner, the image inspection device 200a can inspect the inspection object P for defects.

[0163] In FIG. 24, the operations on the non-defective image from step S38 to step S44 may be performed before the operations on the inspection image from step S31 to step S37, and the operations on the non-defective image and the operations on the inspection image may be performed in parallel.

[0164] The operation of step S45 in Fig. 24 may be performed at any timing from step S31 to step S44. The operation of step S47 may be performed before the operation of step S46, or the operations of step S47 and step S46 may be performed in parallel. The operation of step S49 may be performed before the operation of step S48, or the operations of step S49 and step S48 may be performed in parallel.

[0165] <Processing Results by Processing Unit 260a> Fig. 25 is a diagram showing an example of processed image for second inspection image 62. Fig. 26 is a diagram showing an example of defect candidate area image for inspection image 64 obtained from processed image for second inspection image 62 of Fig. 25.

[0166] In processed image for second inspection image 62 shown in Figure 25, the phase and size of each of the multiple regions obtained by dividing inspection image 60 into a grid are different from those of processed image for first inspection image 61 shown in Figure 7. As a result, processed image for second inspection image 62 is a coarser mosaic-like image compared to processed image for first inspection image 61. Processed image for second inspection image 62 includes defect region 72 corresponding to defect 70 shown in Figure 6 and non-defective feature region 82 corresponding to non-defective feature 80 shown in Figure 6.

[0167] The inspection image defect candidate area image 64 shown in FIG. 26 includes an inspection image defect candidate area 74 corresponding to the defect 70 and an inspection image defect candidate area 84 corresponding to the non-defective feature 80 .

[0168] The effects of the image inspection device 200a other than those described in the second embodiment are similar to those of the image inspection device 200 according to the first embodiment.

[0169] [Third embodiment] Next, an image inspection device according to a third embodiment will be described.

[0170] <Functional configuration of processing unit included in image inspection device according to the third embodiment> 27 is a block diagram showing an example of the functional configuration of a processing unit 260b included in an image inspection device 200b according to the third embodiment. The processing unit 260b has an inspection image divided image generation unit 261b, a non-defective image divided image generation unit 262b, an inspection image defect candidate extraction unit 266b, a non-defective image defect candidate extraction unit 267b, a defect detection unit 268b, and a voting unit 272. This is what makes it different from the processing unit 260a in the second embodiment.

[0171] Each of these parts is a function or means for performing a function that is realized when any of the components shown in FIG. 2 operates in accordance with instructions from CPU 901 in accordance with a program expanded from ROM 902a onto RAM 902b.

[0172] Inspection image divided image generator 261b receives an image signal containing the inspection image captured by imaging unit 210, divides the inspection image into multiple regions, and averages the pixel brightness values ​​in each region to generate N inspection image divided images by varying at least one of the phase, direction, or size of the multiple regions, where N is an integer equal to or greater than 2. Inspection image divided image generator 261b can generate N inspection image divided images with different mosaic coarseness in a mosaic-like image, for example.

[0173] The good-quality image divided image generating unit 262b acquires a good-quality image by referring to the HD 909 or the like, and generates N good-quality image divided images by dividing the good-quality image into a plurality of regions and averaging the luminance values ​​of pixels in each region, while varying at least one of the phase, direction, or size of the plurality of regions. The good-quality image divided image generating unit 262b can generate N good-quality image divided images with different mosaic coarseness in a mosaic-like image, for example.

[0174] In each of the N divided images for inspection image generated by divided image generation unit 261b for inspection image, inspection image defect candidate extraction unit 266b performs a process of comparing the luminance values ​​of pixels in a region of interest for inspection image among multiple regions included in the divided image for inspection image with the luminance values ​​of pixels in a region surrounding the region of interest for inspection image. Inspection image defect candidate extraction unit 266b extracts N defect candidate regions for inspection image based on the first inspection image threshold value generated by first inspection image sensitivity image generation unit 264 and the second inspection image threshold value generated by second inspection image sensitivity image generation unit 270.

[0175] The non-defective image defect candidate extracting unit 267b performs a process of comparing the luminance values ​​of pixels in a region of interest for a non-defective image, among multiple regions included in each of the N non-defective image divided images generated by the non-defective image divided image generating unit 262b, with the luminance values ​​of pixels in a region of interest for a non-defective image surrounding the non-defective image divided image. The non-defective image defect candidate extracting unit 267b extracts N non-defective image defect candidate regions based on the first non-defective image threshold value generated by the first non-defective image sensitivity image generating unit 265 and the second non-defective image threshold value generated by the second non-defective image sensitivity image generating unit 271.

[0176] The defect detection unit 268b compares the N defect candidate areas for the inspection image with the N defect candidate areas for the good product image corresponding to the inspection image defect candidate areas, respectively, to detect N defect areas. For example, the defect detection unit 268b removes (non-extracts) defect candidate areas that are common to both the defect candidate areas for the inspection image and the defect candidate areas for the good product image, assuming that they are characteristics of a good product.

[0177] The defect detection unit 268b can detect areas extracted only from the defect candidate areas for the inspection image as N positive defect areas that are present in the inspection image but not in the good image.Furthermore, the defect detection unit 268b can detect areas extracted only from the defect candidate areas for the good image as N negative defect areas that are present in the good image but not in the inspection image.

[0178] The voting unit 272 generates at least one of a voting defect area for an inspection image obtained by voting N positive defect areas, and a voting defect area for a non-defective image obtained by voting N negative defect areas.

[0179] Specifically, the voting unit 272 prepares a voting space for positive defects and a voting space for negative defects. The voting unit 272 generates a voting defect area for an inspection image by voting N positive defect areas into the voting space for positive defects. The voting unit 272 also generates a voting defect area for a non-defective image by voting N negative defect areas into the voting space for negative defects.

[0180] The voting process is a process in which the first defect area and the second defect area are used as voting source information, respectively, and voting (accumulation) is performed in a voting space set corresponding to the inspection image. The voting space may be configured with a resolution equal to the original resolution of the inspection image and the non-defective image, and voting may be performed for each pixel in the voting space. The voting unit 272 can also perform weighted voting, which weights areas according to a predetermined rule when voting.

[0181] The voting unit 272 outputs at least one of the voting defect area for the inspection image and the voting defect area for the non-defective image as a defect detection result via the output unit 269 to a device or equipment other than the processing unit 260b, such as a PC.

[0182] <Example of operation of image inspection device 200b> Fig. 28 is a flowchart showing an example of the operation of the image inspection device 200b. Fig. 28 shows the operation of the image inspection device 200b, which starts when the inspection object P on which a print image has been formed is sent from the image forming unit 100 to the image inspection device 200b. However, the condition for starting the operation of the image inspection device 200b is not limited to when the inspection object P is sent from the image forming unit 100 to the image inspection device 200b. For example, the image inspection device 200b may start the operation of Fig. 28 when the start condition is an operation input to start inspection made by a user of the image inspection device 200b or the image forming device 1 using the operation panel 101.

[0183] First, in step S61, the image inspection device 200b uses the imaging unit 210 to capture an image of a print image formed on the inspection object P as an inspection image.

[0184] Subsequently, in step S62, the image inspection device 200b causes the divided image for inspection image generating unit 261b to assign 1 to a counter n.

[0185] Next, in step S63, the image inspection device 200b inputs an image signal including the inspection image captured by the imaging unit 210 through the divided image generation unit for inspection image 261b, and divides the inspection image into a plurality of regions that are a matrix lattice.

[0186] Next, in step S64, the image inspection device 200b generates a divided image for inspection image P by the divided image for inspection image generation unit 261b by averaging the luminance values ​​of the pixels included in each of the divided regions. n Generate.

[0187] Next, in step S65, the image inspection device 200b extracts the inspection image segmented image P n The entire defect candidate area U for the inspection image n Extract.

[0188] Subsequently, in step S66, the image inspection device 200b generates the divided image for inspection image P by the divided image for inspection image generating unit 261b, which generates at least one of the phase, direction, and size of each of the divided regions. n , and the brightness values ​​of the pixels included in each region are averaged to obtain a divided image P for the inspection image. n+1 Generate.

[0189] Subsequently, in step S67, the image inspection device 200b determines whether n is equal to N by the divided image for inspection image generating unit 261b.

[0190] If it is determined in step S67 that n is not equal to N (step S67, NO), in step S68, image inspection device 200b causes inspection image divided image generation unit 261b to increment counter n by 1. Thereafter, image inspection device 200b repeatedly performs the operations from step S64 onwards until it is determined in step S67 that n is equal to N.

[0191] On the other hand, if it is determined in step S67 that n is equal to N (step S67, YES), in step S69, the image inspection device 200b acquires a non-defective image by using the non-defective image divided image generation unit 262b with reference to the HD 909 etc.

[0192] Subsequently, in step S70, the image inspection device 200b causes the non-defective divided image generating unit 262b to assign 1 to a counter n.

[0193] Next, in step S71, the image inspection device 200b divides the non-defective image into a plurality of regions that are a matrix lattice by the non-defective image divided image generating unit 262b.

[0194] Next, in step S72, the image inspection device 200b generates a non-defective image divided image Q by the non-defective image divided image generating unit 262b, which averages the luminance values ​​of the pixels included in each of the divided regions. n Generate.

[0195] Subsequently, in step S73, the image inspection device 200b extracts a segmented image Q for a non-defective product image by the defect candidate extraction unit 267b for a non-defective product image. n The defect candidate area for the non-defective image V n Extract.

[0196] Subsequently, in step S74, the image inspection device 200b generates a non-defective image divided image Q by the non-defective image divided image generating unit 262b. The non-defective image divided image Q is generated by dividing at least one of the phase, direction, and size of each of the divided regions. n The brightness values ​​of the pixels in each region are averaged to obtain a segmented image Qn+1 Generate.

[0197] Subsequently, in step S75, the image inspection device 200b determines whether n is equal to N by the non-defective divided image generating unit 262b.

[0198] If it is determined in step S75 that n is not equal to N (step S75, NO), in step S76, the image inspection device 200b causes the divided image generation unit for non-defective images 262b to increment the counter n by 1. Thereafter, the image inspection device 200b repeats the operations from step S72 onwards until it is determined in step S75 that n is equal to N.

[0199] On the other hand, if it is determined in step S75 that n is equal to N (YES in step S75), the defect detection unit 268b of the image inspection device 200b assigns 1 to the counter n in step S77.

[0200] The operations from step S69 to step S76 for the non-defective image may be performed before the operations from step S61 to step S68 for the inspection image, or the two may be performed in parallel.

[0201] Subsequently, in step S78, the image inspection device 200b detects an inspection image defect candidate region U by the defect detection unit 268b. n and the defect candidate area U for the inspection image. n Defect candidate area V for a non-defective image corresponding to n Compare.

[0202] Subsequently, in step S79, the image inspection device 200b detects a positive defect area W by the defect detection unit 268b. n and negative defect region X n Detect.

[0203] Subsequently, in step S80, the image inspection device 200b determines whether n is equal to N by the defect detection unit 268b.

[0204] If it is determined in step S80 that n is not equal to N (step S80, NO), in step S81, the image inspection device 200b causes the defect detection unit 268b to increment the counter n by 1. Thereafter, the image inspection device 200b repeats the operations from step S78 onwards until it is determined in step S80 that n is equal to N.

[0205] On the other hand, if it is determined in step S80 that n is equal to N (step S80, YES), in step S82, the image inspection device 200b uses the voting unit 272 to select N positive defect regions W n The voting defect area for the inspection image obtained by voting is the N negative defect areas X n At least one of the voting defect areas for the non-defective image obtained by voting is generated as a defect detection result.

[0206] Subsequently, in step S83, the image inspection device 200b causes the defect detection unit 268b to output the defect detection result for the printed image via the output unit 269 to a device or equipment other than the processing unit 260b, such as a PC.

[0207] In this way, image inspection device 200b can detect defects in printed images.

[0208] <Processing Results by Processing Unit 260b> Fig. 29 is a diagram showing an example of processing by processing unit 260b, showing an example of processing results for each step in which processing unit 260b inputs inspection image 60 shown in Fig. 6 and non-defective product image 160 shown in Fig. 9 and performs processing.

[0209] Grid-divided images P'1 to P' for inspection images N shows N images obtained by dividing the inspection image 60 by changing the size of the grid-like regions. N shows N images obtained by dividing the non-defective product image 160 by changing the size of the grid-like regions.

[0210] Inspection image divided images P1 to P N are the grid-divided images P'1 to P' for the inspection image. N The figures show N images in which at least one of the phase or direction of the grid-like regions in each of the images is changed. N are the grid-divided images Q'1 to Q' for the non-defective image. N 1 shows N images in which at least one of the phase or orientation of the grid-like regions in each of the images is changed.

[0211] Inspection image defect candidate areas U1 to U N are the divided images P1 to P2 for the inspection image. N The defect candidate areas for the inspection image extracted from each of the defect candidate areas V1 to V2 are shown. N are the divided images Q1 to Q for the good product image. N 10 shows defect candidate areas for non-defective images extracted from each of the images.

[0212] Inspection image defect candidate areas U1 to U N and defect candidate areas U1 to U2 for inspection images. N Defect candidate areas V1 to V2 for non-defective images corresponding to N By comparing the positive defect areas W1 to W N or negative defect areas X1 to X N At least one of the above can be detected.

[0213] Image inspection device 200a shown in the second embodiment corresponds to the case of image inspection device 200b when N=2. Specifically, first processed image for inspection image 61 in Fig. 7 corresponds to divided image for inspection image P1 when n=1, and second processed image for inspection image 62 in Fig. 25 corresponds to divided image for inspection image P2 when n=2. Inspection image defect candidate area image 63 in Fig. 8 corresponds to defect candidate area for inspection image U1. Inspection image defect candidate area image 64 in Fig. 26 corresponds to defect candidate area for inspection image U2.

[0214] Similarly, the first processed image for a non-defective product image 161 in Fig. 10 corresponds to the segmented image for a non-defective product image Q1 when n = 1. The second processed image for a non-defective product image corresponds to the segmented image for a non-defective product image Q2 when n = 2. The defect candidate area image for a non-defective product image 163 in Fig. 11 corresponds to the defect candidate area for a non-defective product image V1.

[0215] The inspection image defect candidate extracting unit 266b extracts N inspection image divided images P1 to P2 including examples of the first inspection image processed image 61 or the second inspection image processed image 62. N and generate N defect candidate areas U1 to U2 for the inspection image including examples of the defect candidate area images 63 and 64 for the inspection image. N Furthermore, the defect candidate extracting unit 267b for non-defective image extracts N divided images Q1 to Q2 for non-defective image including examples of the first processed image for non-defective image 161. N and generate N defect candidate regions V1 to V2 for non-defective images including examples of the defect candidate region image 163 for non-defective images. N Based on these results, the defect detection unit 268b extracts the positive defect areas W1 to W N or negative defect areas X1 to X N At least one of the following is detected.

[0216] Fig. 30 is a diagram showing an example of an inspection image voting defect area 360 in the image inspection device 200b. Fig. 31 is a diagram showing an example of a non-defective image voting defect area 370 in the image inspection device 200b.

[0217] The inspection image voting defect area 360 shown in FIG. 30 is composed of positive defect areas W1 to W N The positive defect 361 is clearly detected.

[0218] The voting defect area 370 for a good image shown in FIG. 31 is the negative defect areas X1 to X N The negative defect 371 is clearly detected.

[0219] <Actions and Effects of Image Inspection Device 200b> As described above, in this embodiment, N positive defect regions W1 to W N 360 for the inspection image obtained by voting the defect area 360, or N negative defect areas X1 to X N At least one of the voting defect areas 370 for the non-defective image is generated by voting the defect area 370.

[0220] By performing the voting process, N positive defect regions W1 to W N In the same way, the N negative defect areas X1 to X2 are detected as positive defects, and the pixel brightness of the positive defect areas is increased compared to other areas. N Within a region, commonly detected negative defects are integrated and have pixel brightness greater than other regions, making them more noticeable. This reduces false detection of defects and enables highly accurate defect detection.

[0221] In this embodiment, N positive defect regions W1 to W N and N negative defect regions X1 to X2 N However, the present invention is not limited to this. N , and select one of the N negative defect regions X1 to X2 as a positive defect detection result. N Alternatively, one of the above may be selected as a negative defect detection result.

[0222] For example, the sizes of the regions may be varied to form N positive defect regions W1 to W2. N and N negative defect regions X1 to X2 N When detecting a positive defect area W, the smaller the area size, the higher the resolution. In this case, for recording media with a rough surface such as plain paper, the surface is likely to be mistakenly detected as a defect area, so the positive defect area W is large in size and has low resolution. n or negative defect area X n By selecting

[0049] , false detections can be effectively suppressed.

[0223] On the other hand, in the case of recording media with a small surface roughness such as glossy paper, the surface is less likely to be mistakenly detected as a defective area, so the size of the area is small and the resolution is high. n or negative defect area X n Selecting increases the resolution and allows for the detection of smaller defects.

[0224] Thus, the positive defect region W n or negative defect area X n By selecting one of the above, it is possible to appropriately perform defect detection according to the characteristics of the object to be inspected.

[0225] The effects of the image inspection device 200b other than those described in the third embodiment are similar to those of the image inspection device 200 according to the first embodiment.

[0226] Although the embodiments have been described above, the present invention is not limited to the specifically disclosed above embodiments, and various modifications and changes are possible without departing from the scope of the claims.

[0227] In the above-described embodiment, a printed image formed by electrophotography is exemplified, but the image inspection device according to the embodiment can also be applied to printed images formed by other methods such as inkjet printing.

[0228] The image inspection device according to the embodiment is a device that performs inspection based on an image of an object to be inspected (inspection image), and the object to be inspected is not limited to an image of a print image, etc. For example, the object to be inspected may be a part, etc.

[0229] The ordinal numbers, quantities, and other numbers used above are all examples for specifically explaining the technology of the present invention, and the present invention is not limited to the exemplified numbers. Furthermore, the connection relationships between the components are examples for specifically explaining the technology of the present invention, and the connection relationships for realizing the functions of the present invention are not limited to these.

[0230] Each function of the above-described embodiments can be realized by one or more processing circuits. Here, the term "processing circuit" in this specification includes a processor programmed to perform each function by software, such as a processor implemented by an electronic circuit, as well as devices such as an ASIC (Application Specific Integrated Circuit), a DSP (Digital Signal Processor), an FPGA (Field Programmable Gate Array), and conventional circuit modules designed to perform each of the above-described functions.

[0231] For example, aspects of the present invention are as follows. <1> an image inspection device for inspecting defects in an object to be inspected, the image pickup unit capturing an inspection image of the object to be inspected; and a processing unit processing the inspection image captured by the image pickup unit, the processing unit including: an inspection image divided image generation unit generating an inspection image divided image using a first inspection image processed image obtained by averaging pixel brightness values ​​in each of a plurality of processing regions obtained by dividing the inspection image according to a predetermined first inspection image division condition; a good product image divided image generation unit generating a good product image divided image using a first good product image processed image obtained by averaging pixel brightness values ​​in each of a plurality of processing regions obtained by dividing a good product image to be compared with the inspection image according to a predetermined first good product image division condition; a sensitivity map image generation unit generating a sensitivity map image of the same size as the inspection image and the good product image, in which a detection threshold is set for each predetermined detection region; and a first inspection image generation unit generating a sensitivity map image by dividing the sensitivity map image into a plurality of regions according to the first inspection image division condition and processing the first inspection image threshold of each region in accordance with the detection threshold. a first inspection image sensitivity image generation unit that divides the sensitivity map image into a plurality of regions in accordance with the first inspection image division condition, and processes the first inspection image threshold value of each region in accordance with the detection threshold value to generate a first inspection image sensitivity image; and a first inspection image sensitivity image generation unit that compares the luminance values ​​of pixels in an inspection image attention region among a plurality of processing regions included in the inspection image division image with the luminance values ​​of pixels in the processing regions surrounding the inspection image attention region, and generates a first inspection image sensitivity image based on the first inspection image threshold value. The image inspection device includes a defect candidate extraction unit for inspection images that extracts defect candidate areas, a defect candidate extraction unit for good images that compares the brightness values ​​of pixels in a region of interest for good images among multiple processing regions included in the divided image for good images with the brightness values ​​of pixels in the processing regions surrounding the region of interest for good images, and extracts defect candidate areas for good images based on the first threshold for good images, and a defect detection unit that compares the defect candidate areas for inspection images with the defect candidate areas for good images and detects defect areas. <2> An image inspection device for inspecting defects in an object to be inspected, the image inspection device comprising: an imaging unit that captures an inspection image of the object to be inspected; and a processing unit that processes the inspection image captured by the imaging unit, wherein the processing unit generates a first processed image for inspection image obtained by averaging the luminance values ​​of pixels in each of a plurality of processing areas obtained by dividing the inspection image under predetermined first division conditions for inspection image; and a second processed image obtained by averaging the luminance values ​​of pixels in each of a plurality of processing areas obtained by dividing the inspection image under second division conditions for inspection image that differ from the first division conditions for inspection image in at least one of phase, direction, and size. a processed image for inspection image generating unit for generating a divided image for inspection image using at least one of a first processed image for non-defective product image obtained by dividing a non-defective product image to be compared with the inspection image under a predetermined first division condition for non-defective product image and a second processed image for non-defective product image obtained by dividing a non-defective product image under a second division condition for non-defective product image, the second processed image being ... a sensitivity map image generation unit for generating a sensitivity map image having the same size as the inspection image and the non-defective image, with a detection threshold set for each predetermined detection region; a sensitivity image generation unit for first inspection image, which divides the sensitivity map image into a plurality of regions according to the division conditions for the first inspection image and generates a sensitivity image for the first inspection image by processing the threshold for the first inspection image included in each region according to the detection threshold; a first-good-quality-image sensitivity image generating unit that generates a first-good-quality-image sensitivity image by processing a threshold value according to the detection threshold value; a second-good-quality-image sensitivity image generating unit that divides the sensitivity map image into a plurality of regions according to the second-good-quality-image dividing condition, and generates a second-good-quality-image sensitivity image by processing a second-good-quality-image threshold value included in each region according to the detection threshold value; and a second-good-quality-image sensitivity image generating unit that divides the sensitivity map image into a plurality of regions according to the second-good-quality-image dividing condition, and generates a second-good-quality-image sensitivity image by processing a second-good-quality-image threshold value included in each region according to the detection threshold value.an inspection image defect candidate extraction unit that compares the brightness values ​​of pixels in an inspection image region of interest among a plurality of processing regions included in the inspection image divided image with the brightness values ​​of pixels in the processing regions around the inspection image region of interest, and extracts an inspection image defect candidate region based on the first inspection image threshold and the second inspection image threshold; a good image defect candidate extraction unit that compares the brightness values ​​of pixels in an inspection image region of interest among a plurality of processing regions included in the good image divided image with the brightness values ​​of pixels in the processing regions around the good image region of interest, and extracts a good image defect candidate region based on the first good image threshold and the second good image threshold; and a defect detection unit that compares the inspection image defect candidate region with the good image defect candidate region to detect a defect region. <3> the detection threshold is set according to any one of the following: a luminance value of the inspection image; a luminance difference between pixels of a region of interest for the inspection image and pixels of the processing region surrounding the region of interest for the inspection image among a plurality of processing regions included in the divided image for the inspection image; or an input by a user of the image inspection device; <1> or the above <2> 2. An image inspection device according to claim 1. <4> The non-defective product image is an image obtained by photographing the non-defective inspection object. <1> From the above <3> The image inspection device is described in any one of the above. <5> The non-defective product image is a digital master image generated based on image data that is the basis of the inspection image. <1> From the above <3> The image inspection device is described in any one of the above. <6> the first inspection image sensitivity image generation unit calculates the first inspection image threshold value by a mathematical formula based on the detection threshold value, and the first non-defective image sensitivity image generation unit calculates the first non-defective image threshold value by a mathematical formula based on the detection threshold value; <1> 2. An image inspection device according to claim 1. <7> the first inspection image sensitivity image generation unit calculates the first inspection image threshold value by a mathematical formula based on the detection threshold value, the first non-defective image sensitivity image generation unit calculates the first non-defective image threshold value by a mathematical formula based on the detection threshold value, the second inspection image sensitivity image generation unit calculates the second inspection image threshold value by a mathematical formula based on the detection threshold value, and the second non-defective image sensitivity image generation unit calculates the second non-defective image threshold value by a mathematical formula based on the detection threshold value; <2> 2. An image inspection device according to claim 1. <8> The aforementioned <1> From the above <7> An image forming apparatus having the image inspection device according to any one of the above. <9> an image inspection method using an image inspection device for inspecting defects in an object to be inspected, the image inspection device capturing an inspection image of the object to be inspected by an imaging unit, and processing the inspection image captured by the imaging unit by a processing unit, the processing unit generating a divided image for inspection image by using a first processed image for inspection image obtained by averaging pixel brightness values ​​in each of a plurality of processing regions obtained by dividing the inspection image according to a predetermined first division condition for the inspection image; generating a divided image for non-defective image by using a first processed image for non-defective image obtained by averaging pixel brightness values ​​in each of a plurality of processing regions obtained by dividing a non-defective image to be compared with the inspection image by using a predetermined first division condition for the non-defective image; generating a sensitivity map image having the same size as the inspection image and the non-defective image, in which a detection threshold is set for each predetermined detection region; and generating a sensitivity map image for first inspection image by using a sensitivity map image generating unit for first inspection image, which is an image of the same size as the inspection image and the non-defective image, in which a detection threshold is set for each predetermined detection region. a first inspection image sensitivity image processed according to the detection threshold is generated; a first non-defective image sensitivity image generation unit divides the sensitivity map image into a plurality of regions according to the first non-defective image division conditions, and generates a first non-defective image sensitivity image by processing the first non-defective image threshold of each region according to the detection threshold; and an inspection image defect candidate extraction unit compares the luminance values ​​of pixels in an inspection image attention region among a plurality of processing regions included in the inspection image division image with the luminance values ​​of pixels in the processing regions around the inspection image attention region. an image inspection method in which a defect candidate area for an inspection image is extracted based on the first threshold value for the inspection image; a defect candidate extraction unit for a good product image compares the brightness values ​​of pixels in a region of interest for a good product image among multiple processing regions included in the divided image for a good product image with the brightness values ​​of pixels in the processing regions surrounding the region of interest for a good product image; a defect candidate area for a good product image is extracted based on the first threshold value for the good product image; and a defect detection unit compares the defect candidate area for an inspection image with the defect candidate area for a good product image to detect a defect area. [Explanation of symbols]

[0232] 1. Image forming device 60 inspection images 61 Processed image for first inspection image 62 Processed image for second inspection image 63, 64 Defect candidate area image for inspection image 70 Defects 71, 72 Defective area 73, 74 Defect candidate area for inspection image 75 Sensitivity Map Images 76 Dead zone 77 Low sensitivity area 78 High sensitivity area 80 Good product characteristics 81, 82 Good product characteristic area 83, 84 Defect candidate area for inspection image 85 Sensitivity image for first inspection image 160 good quality images 161 First processed image for good product image 163 Defect candidate area image for good product image 170 Defects 171 Defective Area 173 Defect candidate area for good product image 180 Good product characteristics 181 Good product characteristic area 183 Defect candidate area for good product image 100 Image forming unit 101 Operation Panel 102 conveying roller pair 103, 103K, 103C, 103M, 103Y Imaging section 104 Fuser roller 105 Transfer belt 106 Reverse Pass 107 Secondary transfer roller 109 Paper feed section 200, 200a, 200b Image inspection device 210 Imaging unit 211, 212, 213 Mirror 214 Lens 215 pixel array 220 Background section 221 Revolver 222 White small diameter roller 223 Black small diameter roller 224 White large diameter roller 225 Black large diameter roller 230 Contact Glass 240 lighting units 250, 250 conveyor roller pair 260, 260a, 260b processing section 261, 261a, 261a inspection image division image generation unit 262, 262a, 262a Division image generation unit for non-defective image 263 Sensitivity map image generation unit 264 Sensitivity image generation unit for first inspection image 265 First good product image sensitivity image generation unit 266, 266b Defect candidate extraction unit for inspection image 267, 267b Defect candidate extraction unit for good product images 268, 268b Defect detection section 269 ​​Output Section 270 Sensitivity image generation unit for second inspection image 271 Sensitivity image generation unit for second good image 272 Voting Department 300 stacker 301 Tray 360 inspection image voting defect area 361 Positive Defect 370 Good Image Voting Defect Area 371 Negative Defects 910 Controller 901 CPU 902 system memory 903 Northbridge 904 Southbridge 906 ASIC 907 Local Memory 908 HDD Controller 909 HD 920 Near field communication circuit 930 Engine control unit 940 Operation Panel 950 Network I / F a Conveyor path P Inspection object s, t arrows [Prior art documents] [Patent documents]

[0233] [Patent Document 1] Patent No. 5821708 [Patent Document 2] Patent No. 5678595

Claims

1. An image inspection device that inspects defects in an inspection object, an imaging unit that captures an inspection image of the inspection object; a processing unit that processes the inspection image captured by the imaging unit, The processing unit a divided image generation unit for an inspection image that generates a divided image for the inspection image using a processed image for a first inspection image obtained by averaging the luminance values ​​of pixels in each of a plurality of processing regions obtained by dividing the inspection image according to predetermined division conditions for the first inspection image; a good-quality image divided image generating unit that generates a good-quality image divided image using a first good-quality image processed image obtained by averaging pixel luminance values ​​in each of a plurality of processing regions obtained by dividing a good-quality image that is to be compared with the inspection image under predetermined first good-quality image dividing conditions; a sensitivity map image generator that generates a sensitivity map image having the same size as the inspection image and the non-defective image, in which a detection threshold is set for each predetermined detection region; a first inspection image sensitivity image generating unit that divides the sensitivity map image into a plurality of regions according to the first inspection image division conditions, and generates a first inspection image sensitivity image by processing the first inspection image threshold value of each region in accordance with the detection threshold value; a first-good-quality-image sensitivity image generating unit that divides the sensitivity map image into a plurality of regions according to the first-good-quality-image dividing condition, and generates a first-good-quality-image sensitivity image by processing the first-good-quality-image threshold value of each region in accordance with the detection threshold value; an inspection image defect candidate extraction unit that compares the brightness values ​​of pixels in an inspection image attention area among a plurality of processing areas included in the inspection image divided image with the brightness values ​​of pixels in the processing areas surrounding the inspection image attention area, and extracts an inspection image defect candidate area based on the first inspection image threshold; a defect candidate extracting unit for a non-defective image that compares the brightness values ​​of pixels in a region of interest for a non-defective image among a plurality of processing regions included in the divided image for a non-defective image with the brightness values ​​of pixels in the processing regions around the region of interest for a non-defective image, and extracts a defect candidate region for a non-defective image based on the first threshold value for a non-defective image; and a defect detection unit that compares the defect candidate areas for the inspection image with the defect candidate areas for the non-defective product image to detect defect areas.

2. An image inspection device that inspects defects in an inspection object, an imaging unit that captures an inspection image of the inspection object; a processing unit that processes the inspection image captured by the imaging unit, The processing unit a division image generation unit for an inspection image that generates a division image for the inspection image using at least one of a first processed image for the inspection image obtained by averaging the brightness values ​​of pixels in each of a plurality of processing areas obtained by dividing the inspection image under predetermined division conditions for the first inspection image, and a second processed image for the inspection image obtained by averaging the brightness values ​​of pixels in each of a plurality of processing areas obtained by dividing the inspection image under second division conditions for the inspection image that differ from the division conditions for the first inspection image in at least one of phase, direction, and size; a good-quality image divided image generation unit that generates a good-quality image divided image using at least one of a first good-quality image processed image obtained by averaging the luminance values ​​of pixels in each of a plurality of processing regions obtained by dividing a good-quality image, which is a comparison target of the inspection image, under a predetermined first good-quality image division condition, and a second good-quality image processed image obtained by averaging the luminance values ​​of pixels in each of a plurality of processing regions obtained by dividing the good-quality image under a second good-quality image division condition that differs from the first good-quality image division condition in at least one of phase, direction, and size; a sensitivity map image generator that generates a sensitivity map image having the same size as the inspection image and the non-defective image, in which a detection threshold is set for each predetermined detection region; a first inspection image sensitivity image generation unit that divides the sensitivity map image into a plurality of regions according to the first inspection image division conditions, and generates a first inspection image sensitivity image by processing the first inspection image threshold value included in each region in accordance with the detection threshold value; a first-good-product-image sensitivity image generating unit that divides the sensitivity map image into a plurality of regions according to the first-good-product-image dividing condition, and generates a first-good-product-image sensitivity image by processing the first-good-product-image threshold value included in each region in accordance with the detection threshold value; a second inspection image sensitivity image generation unit that divides the sensitivity map image into a plurality of regions according to the second inspection image division conditions, and generates a second inspection image sensitivity image by processing the second inspection image threshold value included in each region in accordance with the detection threshold value; a second-good-quality-image sensitivity image generating unit that divides the sensitivity map image into a plurality of regions according to the second-good-quality-image dividing condition, and generates a second-good-quality-image sensitivity image by processing the second-good-quality-image threshold value included in each region in accordance with the detection threshold value; an inspection image defect candidate extraction unit that compares the brightness values ​​of pixels in an inspection image region of interest among a plurality of processing regions included in the inspection image divided image with the brightness values ​​of pixels in the processing regions surrounding the inspection image region of interest, and extracts an inspection image defect candidate region based on the first inspection image threshold and the second inspection image threshold; a defect candidate extracting unit for a good product image that compares the brightness values ​​of pixels in a region of interest for a good product image among a plurality of processing regions included in the divided image for a good product image with the brightness values ​​of pixels in the processing regions around the region of interest for a good product image, and extracts a defect candidate region for a good product image based on the first threshold value for a good product image and the second threshold value for a good product image; and a defect detection unit that compares the defect candidate areas for the inspection image with the defect candidate areas for the non-defective product image to detect defect areas.

3. The image inspection device described in claim 1 or claim 2, wherein the detection threshold is set according to one of the following: the brightness value of the inspection image; the brightness difference between the pixels of the region of interest for the inspection image and the pixels of the processing region surrounding the region of interest for the inspection image among multiple processing regions included in the divided image for the inspection image; or input by a user of the image inspection device.

4. 3. The image inspection device according to claim 1, wherein the non-defective product image is an image obtained by photographing the non-defective inspection object.

5. 3. The image inspection device according to claim 1, wherein the non-defective product image is a digital master image generated based on image data that is the basis of the inspection image.

6. the first inspection image sensitivity image generation unit calculates the first inspection image threshold value by a mathematical formula based on the detection threshold value; The image inspection device according to claim 1 , wherein the first non-defective image sensitivity image generating unit calculates the first non-defective image threshold value by a mathematical formula based on the detection threshold value.

7. the first inspection image sensitivity image generation unit calculates the first inspection image threshold value by a mathematical formula based on the detection threshold value; the first non-defective image sensitivity image generation unit calculates the first non-defective image threshold value by a mathematical formula based on the detection threshold value; the second inspection image sensitivity image generation unit calculates the second inspection image threshold value by a mathematical formula based on the detection threshold value; The image inspection device according to claim 2 , wherein the second non-defective image sensitivity image generating unit calculates the second non-defective image threshold value by a mathematical formula based on the detection threshold value.

8. An image forming apparatus comprising the image inspection device according to claim 1 or 2.

9. An image inspection method using an image inspection device that inspects an object for defects, comprising: The image inspection device An inspection image of the inspection object is captured by an imaging unit; a processing unit processes the inspection image captured by the imaging unit; The processing unit a divided image generation unit for the inspection image generates a divided image for the inspection image using a first processed image for the inspection image obtained by averaging the luminance values ​​of pixels in each of a plurality of processing regions obtained by dividing the inspection image according to a predetermined first division condition for the inspection image; a good-quality image divided image generating unit generates a good-quality image divided image using a first good-quality image processed image obtained by averaging pixel luminance values ​​in each of a plurality of processing regions obtained by dividing the good-quality image to be compared with the inspection image under predetermined first good-quality image dividing conditions; a sensitivity map image generating unit generating a sensitivity map image having the same size as the inspection image and the non-defective image, the sensitivity map image having a detection threshold set for each predetermined detection region; a first inspection image sensitivity image generating unit for dividing the sensitivity map image into a plurality of regions according to the first inspection image division conditions, and generating a first inspection image sensitivity image by processing the first inspection image threshold value of each region in accordance with the detection threshold value; a first-good-product-image sensitivity image generating unit divides the sensitivity map image into a plurality of regions according to the first-good-product-image division conditions, and generates a first-good-product-image sensitivity image by processing the first-good-product-image threshold value of each region in accordance with the detection threshold value; an inspection image defect candidate extraction unit compares the brightness values ​​of pixels in an inspection image attention area among a plurality of processing areas included in the inspection image divided image with the brightness values ​​of pixels in the processing areas surrounding the inspection image attention area, and extracts the inspection image defect candidate area based on the first inspection image threshold; a defect candidate extraction unit for non-defective image compares the brightness values ​​of pixels in a region of interest for a non-defective image among a plurality of processing regions included in the divided image for a non-defective image with the brightness values ​​of pixels in the processing regions around the region of interest for a non-defective image, and extracts a defect candidate region for a non-defective image based on the first threshold value for a non-defective image; an image inspection method, wherein a defect detection unit compares the defect candidate area for the inspection image with the defect candidate area for the non-defective image to detect a defect area;

Images