JP2021071301A - Image inspection device and image inspection system - Google Patents

Abstract

To solve the problem that streak-like defects differing in length cannot be detected with good accuracy.SOLUTION: An image inspection device 3 comprises: a change amount calculation unit 611 for calculating a classification result obtained by classifying a difference value between a pixel of interest included in a read image and a comparison pixel by a classification threshold into a plurality of classification values, as a change amount of pixel value; a defect feature quantity extraction unit 612 for extracting, for each first defect detection region whose length in a sub-scan direction defines a first size, a first feature quantity of a streak-like defect occurring in an image formed on a recording material from the classification result, and extracting, for each second defect detection region whose length in the sub-scan direction defines a second size shorter then the first size, a second feature quantity of a streak-like defect from the classification result; and a quality determination unit 613 for comparing the first feature quantity with a first defect detection threshold and comparing the second feature quantity with a second defect detection threshold higher than the first defect detection threshold, detecting streak-like defects of different lengths in the sub-scan direction, and determining the quality of the image.SELECTED DRAWING: Figure 4

Description

The present invention relates to an image inspection device and an image inspection system.

When printing a large amount of prints, the operator usually performs trial printing several times to make various adjustments, and when it is confirmed by a sample that there is no problem in the printed contents, the main printing is started. However, in the image of the printed paper, color deviation or distortion may occur with respect to the sample for some reason. Therefore, in recent years, there has been provided an inspection system in which a scanning device such as a scanner is provided on a transport path in the subsequent stage of the image forming apparatus, and the image inspection device inspects the scanned image obtained by reading the image of each paper to be printed out. It became so. In this inspection, an image that is the source of the image formed on the paper is saved as an output target image. Then, the image inspection device compares the image (scanned image) obtained by scanning the paper printed by the main printing with a scanner and the stored output target image.

Comparing the scanned image with the output target image, image defects may be detected in the scanned image. Examples of the types of image defects include narrow streaks and bands that appear continuously in a predetermined direction or interrupted in a predetermined direction. Such image defects are collectively referred to as streak-like defects. However, in the following description, the streak-like defect is also abbreviated as "streak". A streak is a defect caused by scratches or stains on a drum or roller of a reading device such as an image forming device or a scanner, and is a gradation difference (also referred to as a brightness value) of a pixel value (also referred to as a brightness value) not intended by the operator. It also appears as "streak strength") or uneven density. Examples of the streaks include white streaks that are lighter than the original image, black streaks that are darker than the original image, and the like. Since streaks are likely to occur repeatedly and are easily recognized by operators and customers, it is required to reliably detect streaks even during the main printing.

A technique disclosed in Patent Document 1 is known for detecting streaks. In Patent Document 1, a technique of creating a composite image obtained by synthesizing a plurality of low-resolution images having different resolutions created while changing the size of a streak detection area set in the captured image, and extracting defect candidates from the composite image. Is disclosed.

Japanese Unexamined Patent Publication No. 2019-100937

The streaks appear in the sub-scanning direction or the main scanning direction of the image formed on the paper due to various factors. For example, when the band electrode included in the image forming apparatus is contaminated with toner or the like, the contaminated portion may not be charged well, and as a result, it may appear as a streak in the sub-scanning direction of the image. Further, when a defect occurs in a rotating body such as a photoconductor drum, streaks appear in the main scanning direction in synchronization with the rotation cycle of the rotating body. Here, the streaks appearing in the scanned image will be described with reference to FIG.

FIG. 1 is a graph showing an example of data in which pixel values are averaged in the sub-scanning direction (FD: Feed direction). The horizontal axis of this graph represents the main scanning position of the scanned image in the main scanning direction (CD: Cross direction), and the vertical axis represents the pixel value for each pixel of the scanned image read in the G-ch (green channel). Represent. This graph is an average of the pixel values of the pixels in the sub-scanning direction at the same position in the main scanning direction based on the result of reading a flat image by a scanning device such as a scanner. In the following example, the printing direction in which images and characters are formed on the sheet will be described as matching the main scanning direction, but the same applies when the printing direction coincides with the sub-scanning direction. It is assumed that the scanned image shown in FIG. 1 has two streaks in the sub-scanning direction.

Conventionally, in order to detect streaks generated in a scanned image, a method of searching for streaks by setting one type of detection threshold value for a streak detection area divided into a certain length has been adopted. However, in this method, false detection of streaks such as detecting noise that does not correspond to the streak detection region as streaks occurs, and the streak detection accuracy is deteriorated. The technique disclosed in Patent Document 1 divides the scanned image into a finite number of regions and performs streak detection from the streak detection region. However, since there is no description of changing the detection level (threshold value for the feature amount) corresponding to the inspection (division) region, the streak detectability of different lengths is inferior.

The present invention has been made in view of such a situation, and an object of the present invention is to enable detection of streak-like defects having different lengths.

In order to achieve at least one of the above-mentioned objects, an image inspection apparatus reflecting one aspect of the present invention is noted among a plurality of pixels included in a scanned image read from a recording material on which an image is formed. The difference value of the pixel value of the comparison pixel separated by the number of the first pixel in the first direction with respect to the pixel value of the pixel with respect to the attention pixel is calculated, and the difference value is classified into a plurality of classification values by the classification threshold. For each of the change amount calculation unit that calculates the change amount of the pixel value and the first defect detection region having the length of the second direction intersecting the first direction of the classification result as the first size, the recording material. A second defect in which the first feature amount of the streak-like defect generated in the image formed in the image is extracted from the classification result and the length in the second direction is set to the second size shorter than the first size. For each detection region, the feature amount extraction unit that extracts the second feature amount of the streak-like defect from the classification result, the first feature amount is compared with the first defect detection threshold, and the second feature amount is the first. With a quality judgment unit that detects streaky defects of different lengths in the second direction and judges the quality of the image formed on the recording material as compared with the second defect detection threshold that is higher than the defect detection threshold of , Equipped with.
The above-mentioned image inspection apparatus is one aspect of the present invention, and an image inspection system reflecting one aspect of the present invention is configured in the same manner as the above-mentioned image inspection apparatus.

According to the present invention, since the streak-like defects of different lengths are detected with high accuracy, the quality of the image formed on the recording material can be improved.
Issues, configurations and effects other than those described above will be clarified by the following description of the embodiments.

It is a graph which shows an example of the data which averaged the pixel value in the sub-scanning direction. It is a schematic block diagram of the image inspection system which concerns on 1st Embodiment of this invention. It is a block diagram which shows the structural example of the control system of the image forming apparatus which concerns on 1st Embodiment of this invention. It is a block diagram which shows the structural example of the control system of the image inspection apparatus which concerns on 1st Embodiment of this invention. It is a figure which shows the example of the scanned image which contains the streak to be detected which concerns on 1st Embodiment of this invention, and the classification result. It is a figure which shows the state of extracting the streak feature amount from the classification result which concerns on 1st Embodiment of this invention. It is a figure which shows a mode that the change amount calculation part which concerns on 1st Embodiment of this invention lowers the resolution in the sub-scanning direction, and calculates the ternary average value. It is a figure which shows the example of the ternary pixel included in the classification result which concerns on 1st Embodiment of this invention, and the position of a noise, a short streak and a long streak appearing in a classification result. It is a figure which shows a mode that the defect feature amount extraction part which concerns on 1st Embodiment of this invention extracts a streak feature amount of a long streak from a ternation average value of a long streak. It is a figure which shows a mode that the defect feature amount extraction part which concerns on 1st Embodiment of this invention extracts a noise and a streak feature amount of a short streak from a ternation average value of a short streak. It is a flowchart which shows the example of the processing performed by the inspection processing apparatus which concerns on 1st Embodiment of this invention. It is a figure which shows the example in which a part of the plurality of streak detection regions which concern on the modification of 1st Embodiment of this invention overlap. It is a block diagram which shows the structural example of the control system of the image inspection apparatus which concerns on 2nd Embodiment of this invention. It is a graph which shows the example of the pixel value and the difference value of the pixel in the flat region and the non-flat region of the scanned image which concerns on 2nd Embodiment of this invention, and the classification threshold value. It is a flowchart which shows the example of the processing performed by the inspection processing apparatus which concerns on 2nd Embodiment of this invention.

Hereinafter, embodiments for carrying out the present invention will be described with reference to the accompanying drawings. In the present specification and the drawings, components having substantially the same function or configuration are designated by the same reference numerals, and redundant description will be omitted.

[First Embodiment]
<Configuration of image inspection system>
By taking the difference between the pixel of interest included in the scanned image and the comparison pixel near the pixel of interest, the present inventor is inseparable from the noise as shown in FIG. 1 and is visible to the operator. We have invented an image inspection device that detects streaks of different lengths. Hereinafter, a configuration example and an operation example of an image inspection system including an image inspection device capable of detecting streaks from a scanned image according to the first embodiment will be described with reference to FIGS. 2 to 11.

First, a configuration example of the image inspection system according to the first embodiment of the present invention will be described with reference to FIG.
FIG. 2 is a schematic configuration diagram of the image inspection system 1 according to the first embodiment of the present invention. Note that FIG. 2 shows elements considered necessary for the description of the present invention or related elements thereof, and the image inspection system of the present invention is not limited to the example shown in FIG.

The image inspection system 1 includes an image forming device 2 and an image inspection device 3. The image forming apparatus 2 is an example of an image forming apparatus that forms an image on paper by an electrophotographic method that forms an image using static electricity. The image forming apparatus 2 forms a color image on paper in a tandem format in which toner images of four colors of yellow (Y), magenta (M), cyan (C), and black (K) are superimposed, for example. A PC (Personal Computer) 6 (see FIG. 3 to be described later) operated by an operator is connected to the image forming apparatus 2 via a LAN (Local Area Network) (not shown). Then, a job is input from the PC 6 to the image forming apparatus 2 via the LAN. The image forming apparatus 2 performs various processes such as an image forming process according to the submitted job.

First, a configuration example of the image forming apparatus 2 will be described.
The image forming apparatus 2 includes an image input unit 11 and an operation display unit 13 having an automatic document feeder (ADF: Auto Document Feeder) 12. Further, the image forming apparatus 2 includes a printer unit 10 having a paper feed tray 20 and an image forming unit 30.

The image input unit 11 optically reads an image from the document on the platen of the ADF 12 and A / D-converts the read image to generate image data. The image input unit 11 can also read an image from the original on the platen glass.

The operation display unit 13 is composed of a display unit including a liquid crystal panel and the like, and an operation unit including a touch sensor and the like. The display unit and the operation unit are integrally formed as, for example, a touch panel. The operation display unit 13 generates an operation signal indicating the operation content from the operator input to the operation unit, and supplies the operation signal to the control unit 50 (see FIG. 3 to be described later). Further, the operation display unit 13 displays the operation content and setting information by the operator on the display unit based on the display signal supplied from the control unit 50. It is also possible to configure the operation unit with a mouse, tablet, or the like, and to configure it separately from the display unit.

The paper feed tray 20 is a container for accommodating the paper Sh on which the image is formed by the image forming unit 30. Papers having different paper types, basis weights, and the like are stored in the paper feed trays 20. Paper Sh is an example of a recording material. The image forming apparatus 2 can also form an image on a resin sheet which is an example of a recording material. In the present embodiment, an example in which two paper feed trays 20 are provided has been given, but the number of paper feed trays 20 may be one or three or more.

The image forming apparatus 2 is provided with a transport path 21 for transporting the paper Sh fed from the paper feed tray 20 to the image inspection device 3. The transport path 21 is provided with a plurality of transport rollers for transporting the paper Sh.

On the downstream side of the fixing portion 36, the transport path 21 extends and is connected to the transport path 41 of the image inspection device 3. Further, the transport path 21 branches on the downstream side of the fixing portion 36. An inverted transport path 22 that joins the transport path 21 on the upstream side of the printer unit 10 is connected to one end of the branched transport path 21. The reversing transfer path 22 is provided with a reversing portion 23 for reversing the paper Sh. The paper Sh flipped by the reversing section 23 is returned to the upstream side of the transport path 21 through the reverse transport path 22. Further, the paper Sh that has been inverted due to the route switching may be returned to the transport path 21 on the downstream side of the fixing portion 36 and then transported to the image inspection device 3.

The image forming unit 30 includes four image forming units 31Y, 31M, 31C and 31K for forming toner images of each color of Y, M, C and K, and forms an image on paper Sh. The image forming units 31Y, 31M, 31C and 31K have a charging unit, an exposure unit (all not shown), a photoconductor drum 32Y, 32M, 32C, 32K as an image carrier, and a developing unit 33Y, 33M, 33C, respectively. , 33K.

The developing units 33Y, 33M, 33C, and 33K irradiate the surfaces (outer peripheral portions) of the photoconductor drums 32Y, 32M, 32C, and 32K with light according to the image, thereby electrostatically irradiating the peripheral portions of the photosensitizing drums. Form a latent image. Then, the developing units 33Y, 33M, 33C, 33K form a toner image on the photoconductor drums 32Y, 32M, 32C, 32K by adhering the toner to the electrostatic latent image.

Further, the image forming unit 30 includes an intermediate transfer belt 34, a secondary transfer unit 35, and a fixing unit 36. The intermediate transfer belt 34 is a belt on which images formed on the photoconductor drums 32Y, 32M, 32C, and 32K are primarily transferred. The secondary transfer unit 35 is a roller that secondarily transfers the toner images of each color primary transferred onto the intermediate transfer belt 34 onto the paper Sh transported along the transport path 21.

The fixing unit 36 is arranged on the downstream side of the secondary transfer unit 35 in the paper transport direction, and performs a fixing process on the paper Sh on which the color toner image supplied from the image forming unit 30 is formed. The fixing unit 36 fixes the image transferred by the image forming unit 30 on the surface side of the paper Sh by heating and pressurizing the conveyed paper Sh. The paper Sh on which the image is fixed by the fixing section 36 is conveyed to the image inspection device 3 by the conveying path 21, or is conveyed on the upstream side of the printer section 10 after being inverted by the inversion section 23 through the inversion conveying path 22. Returned to Road 21. An image is formed on the back side of the paper Sh whose front and back sides are reversed by the printer unit 10. After that, the paper Sh that has been fixed by the fixing unit 36 is conveyed to the image inspection device 3.

Next, a configuration example of the image inspection device 3 will be described.
The image inspection device 3 performs an image inspection for detecting streaks generated in an image formed (printed) on the paper Sh conveyed from the image forming device 2. The processing of the image formed on the paper Sh, that is, the inspection of the image by the image inspection device 3, is mainly performed by the inspection processing device 5 attached to the image inspection device 3.

The image inspection device 3 is conveyed on the transfer paths 41, 42, 43, the switching unit 44, the reading units 45a, 45b, the colorimeter 46, and the transfer path 41 that convey the paper Sh conveyed from the image forming apparatus 2. It has paper ejection trays 47 and 48 from which the paper Sh is ejected.

The reading units 45a and 45b are examples of image input devices such as image sensors, respectively. The reading units 45a and 45b project light onto the surface of the paper Sh, for example, and capture the reflected light from the paper Sh as image data. The fact that the reading units 45a and 45b capture the image data of the paper Sh in this way is called "reading". The reading unit 45a reads the paper Sh transported through the transport path 41 from below the transport path 41, and the reading unit 45b reads the paper Sh conveyed through the transport path 41 from above the transport path 41. In the following description, since the reading units 45a and 45b are not distinguished, they are collectively referred to as "reading unit 45". Then, the reading unit 45 outputs the captured image data to the inspection processing device 5.

The colorimeter 46 reads an image formed on the upper surface of the paper Sh transported through the transport path 41, and measures the color density (reflection density) of the image based on the image information obtained by reading the image. This is an example of a device. The colorimeter 46 is, for example, a colorimeter capable of measuring the intensity (spectrum) of reflected light for each wavelength of light, and measures the measured color density (reflection density), L * a * b * value, and the like. Output. For the colorimeter 46, for example, a scanner (line sensor) in which a plurality of sensors (photoelectric conversion elements) (not shown) are arranged one-dimensionally over the entire area in the paper width direction (direction orthogonal to the paper transport direction) is used. To. When the colorimeter 46 is configured by a scanner, the image is read while moving the scanner in a direction orthogonal to the arrangement direction (paper transport direction). Then, the colorimeter 46 measures the color density of the image formed on the paper Sh for each region obtained by dividing the region where the image is read into a mesh shape. The colorimeter 46 outputs the measured color density information to the inspection processing device 5.

The colorimeter 46 may be composed of a single sensor, and the sensor may be moved two-dimensionally to measure the color density of the image formed on the paper Sh. Alternatively, the colorimeter 46 may be composed of a plurality of sensors arranged two-dimensionally (matrixally), and the plurality of sensors may read the color density of all pixels on the paper by one measurement.

The image inspection device 3 includes transport paths 42 and 43 connected to the transport path 41.
The transport path 42 is a path that branches from the middle of the transport path 41, and the paper Sh inspected by the inspection processing device 5 is discharged to the paper discharge tray 47 (an example of the paper discharge unit). Paper Sh (also referred to as “normal paper”) whose image is determined to be normal by the inspection processing device 5 is discharged to the paper ejection tray 47.

The transport path 43 is also a path that branches from the middle of the transport path 41, and the paper Sh inspected by the inspection processing device 5 is discharged to the paper discharge tray 48 (an example of the paper discharge portion). Paper Sh (also referred to as “abnormal paper”) whose image is determined to be abnormal by the inspection processing device 5 is discharged to the paper ejection tray 48.

The switching unit 44 switches the transport direction of the paper Sh so that the paper Sh is transported to either the transport path 42 or 43. If the image inspection device 3 has only one paper ejection tray 47, normal paper and abnormal paper are mixed and ejected. In this case, the normal paper and the abnormal paper are discharged, for example, with a slight shift in the direction orthogonal to the discharge direction.

The paper Sh conveyed to the image inspection device 3 is a printed matter having an image formed on both sides or one side. The image inspection device 3 reads the image formed on both sides or one side of the paper Sh by the image forming device 2, and the inspection processing device 5 performs a predetermined inspection.

In this embodiment, since the image forming apparatus 2 can form an image on both sides of the paper Sh, the inspection processing apparatus 5 inspects both sides of the paper Sh. However, the inspection processing device 5 may be configured to inspect only one side of the paper Sh conveyed from the image forming apparatus capable of forming an image on only one side of the paper Sh.

[Structure of control system of image forming apparatus]
Next, a configuration example of the control system of the image forming apparatus 2 will be described with reference to FIG.
FIG. 3 is a block diagram showing a configuration example of the control system of the image forming apparatus 2.
The image forming apparatus 2 mainly includes a communication I / F unit 51, a paper conveying unit 24, an image input unit 11, an image forming unit 30, a control unit 50, a storage unit 52, a fixing unit 36, and an operation display unit 13. Be prepared.

The communication I / F unit 51 is an interface for transmitting and receiving data to and from a PC 6 which is a terminal operated by an operator via a network or a dedicated line. As the communication I / F unit 51, for example, a NIC (Network Interface Card) is used.

The paper transport unit 24 drives a transport path 21 shown in FIG. 2, a transport roller (not shown) provided on the reverse transport path 22, and a reverse section 23 based on control by the control unit 50.

The control unit 50 includes a CPU (Central Processing Unit) 501, a ROM (Read Only Memory) 502, a RAM (Random Access Memory) 503, and an input image processing unit 504.
The ROM 502 stores a program executed by the CPU 501 of the control unit 50, data used when executing the program, and the like. The CPU 501 controls each part constituting the image forming apparatus 2 by reading the program stored in the ROM 502.
Variables and parameters generated during the arithmetic processing of the CPU 501 are temporarily written in the RAM 503.

The input image processing unit 504 performs predetermined image processing (for example, rasterization processing) on the input image included in the job input from the PC 6 via the communication I / F unit 51 to create print image data. Further, the input image processing unit 504 also performs image processing on the image data acquired from the document read by the image input unit 11 by the ADF 12 or the image data acquired from the outside to create image data for printing. The image data for printing is sent to the image forming unit 30 and the image inspection device 3. In the image inspection apparatus 3, the image data for printing is stored as an output target image 603b (see FIG. 4 described later).

The control unit 50 controls the paper transport unit 24 to drive the transport roller to transport the paper Sh on the transport path 21. Further, the control unit 50 outputs the print image data created by the input image processing unit 504 to the image forming unit 30. Further, the control unit 50 controls the image forming unit 30 to form an image on the paper Sh. Further, the control unit 50 controls the fixing unit 36 to fix the image on the paper Sh.

Further, the control unit 50 receives an operation signal from the operation display unit 13 and performs control according to the operation signal. Further, the control unit 50 outputs a display signal to the operation display unit 13, and the operation display unit 13 displays an operation screen for displaying various setting screens for inputting various operation instructions and setting information, various processing results, and the like. Display on the panel. The information displayed on the operation display unit 13 also includes the streak detection result 631 (see FIG. 4 described later) output from the image inspection device 3.

The storage unit 52 stores parameters used when the CPU 501 of the control unit 50 executes the program, data obtained by executing the program, and the like. For example, the storage unit 52 stores information such as image formation conditions for each density level. The storage unit 52 may store the program executed by the CPU 501.

[Configuration of control system for image inspection equipment]
Next, a configuration example of the control system of the image inspection device 3 will be described with reference to FIG.
FIG. 4 is a block diagram showing a configuration example of the control system of the image inspection device 3.

The image inspection device 3 includes a communication I / F unit 61, a paper transport unit 62, a reading unit 45, and a colorimeter 46 as main configurations. Further, the inspection processing device 5 attached to the image inspection device 3 includes a control unit 60 and a storage unit 63. Further, a storage device 4 is attached to the inspection processing device 5.

The communication I / F unit 61 is an interface for transmitting and receiving data to and from the image forming apparatus 2 via a network. For example, NIC is used as the communication I / F unit 61.
The paper transport unit 62 drives a transport roller (not shown) provided on the transport path 41 shown in FIG. 2 based on the control by the control unit 60.

As described above, the reading unit 45 reads the images formed on the upper surface and the lower surface of the paper Sh transported through the transport path 41. In the present embodiment, the image data of the paper Sh read by the reading units 45a and 45b is referred to as a "read image".

The image read by the reading unit 45 from the paper Sh on which the image is formed is stored in the RAM 603 of the control unit 60 as the read image 603a. Further, the RIP-processed print image data received by the inspection processing device 5 from the image forming device 2 is stored in the RAM 603 as an output target image 603b. As will be described in the second embodiment described later, the output target image 603b may also be used for inspecting the read image 603a. The scanned image 603a and the output target image 603b may be stored in a storage unit 63 composed of a large-capacity HDD or the like. Further, the color density information output from the colorimeter 46 to the image inspection device 3 may be included in the scanned image 603a and the output target image 603b.

The control unit 60 includes a CPU 601 and a ROM 602, a RAM 603, a change amount calculation unit 611, a defect feature amount extraction unit 612, and a quality determination unit 613.

The CPU 601 controls each part constituting the image inspection device 3 by reading the program stored in the ROM 602. By executing the program read from the ROM 602 by the CPU 601, the functions of the change amount calculation unit 611, the defect feature amount extraction unit 612, and the quality determination unit 613 are realized.

The ROM 602 stores a program executed by the CPU 601 of the control unit 60, data used when executing the program, and the like. The ROM 602 is used as an example of a computer-readable, non-transient recording medium that stores a program executed by the CPU 601.
Variables and parameters generated during the arithmetic processing of the CPU 601 are temporarily written in the RAM 603. As described above, the read image 603a, the output target image 603b, the difference image 603c1, the classification result 603d, and the parameter 603e are also stored in the RAM 603.

The parameter 603e includes various values set by the control unit 60 or the operator. The parameter 603e includes, for example, a defect detection threshold value for detecting a streak from a streak feature amount described later. The change amount calculation unit 611, the defect feature amount extraction unit 612, and the quality determination unit 613 perform various processes based on various values read from the parameter 603e.

The change amount calculation unit 611 is a pixel of the comparison pixel 72 that is separated by the first pixel number in the first direction with respect to the pixel value of the attention pixel 71 with respect to the pixel value of the attention pixel 71 among the plurality of pixels included in the scanned image 603a. The difference value of the values is calculated, and the classification result 603d in which the difference value is classified into a plurality of classification values by the classification threshold is calculated as the amount of change in the pixel value. Examples of the attention pixel 71 and the comparison pixel 72 are shown in FIG. 5, which will be described later. Hereinafter, the first direction will be described as the main scanning direction, and the second direction will be described as the sub-scanning direction. However, if the direction of the streaks to be detected is the main scanning direction, the first direction may be the sub-scanning direction and the second direction may be the main scanning direction. That is, the first direction is either a horizontal direction parallel to the printing direction or a vertical direction perpendicular to the printing direction. Further, in the present embodiment, the number of first pixels is set to "3".

For example, the change amount calculation unit 611 calculates the difference value between the pixel value of the attention pixel 71 selected from the plurality of pixels included in the scanned image 603a and the pixel value of the comparison pixel 72 as the change amount. The comparison pixel 72 is a pixel located at a position separated by the number of first pixels in the first direction with respect to the pixel of interest 71. Then, the change amount calculation unit 611 generates a difference image 603c1 including the difference value calculated by performing the same difference processing on the other pixels of the read image 603a. At this time, the change amount calculation unit 611 takes a difference between the shift image (not shown) obtained by shifting the read image 603a in the horizontal or vertical direction by a predetermined number of pixels and the read image 603a before the shift, and obtains a difference image. It is possible to generate 603c1. The shift represents the distance between the pixels for determining the comparison pixel 72 with respect to the pixel of interest 71.

Further, the change amount calculation unit 611 generates a classification result 603d classified into a plurality of pixels for each pixel as a change amount based on the size and code of the difference value of each pixel included in the difference image 603c1. Here, the classification result 603d is represented as a result of the change amount calculation unit 611 classifying (trivalentizing) the difference value by the classification threshold value for each of a plurality of pixels included in the difference image 603c1 (FIG. 5 described later). Lower row). The classification threshold is a value set in the parameter 603e.

In the present embodiment, the process in which the change amount calculation unit 611 refers to the classification threshold value and digitizes the difference value of each pixel included in the difference image 603c1 for each pixel is referred to as “multi-valued”. Depending on the classification threshold value, the change amount calculation unit 611 may multi-value the difference value of the difference image 603c1 to binarize, quaternize, or the like. Here, the classification result 603d is represented as a result of the change amount calculation unit 611 classifying (trivalentizing) the difference value by the classification threshold value for each of a plurality of pixels included in the difference image 603c1 (FIG. 5 described later). Lower row).

The classification threshold is a value set in the parameter 603e. The classification threshold value is a value set in advance before the start of the image inspection, and may be set to a different value depending on the resolution of the scanned image 603a, the type of the image formed on the paper Sh, and the like. Details of the classification result 603d will be described with reference to FIGS. 7 and 8 described later.

The defect feature amount extraction unit 612 is a feature amount of streak-like defects generated in an image formed on the paper Sh based on a change amount calculated based on a plurality of pixels in the sub-scanning direction (“streak feature amount”). Is called) is extracted. For example, when the direction in which the streaks are generated is the sub-scanning direction, the defect feature amount extraction unit 612 refers to the classification result 603d for each predetermined position in the main scanning direction in the sub-scanning direction, and refers to the averaging result (three values). Extract the streak features from the averaging result). The ternary averaging result is a result obtained by the defect feature amount extraction unit 612 averaging the difference values classified by the classification threshold value by the change amount calculation unit 611 and multi-valued. It is shown in the graph represented by the ternary average value shown in the middle row.

In the present embodiment, the defect feature amount extraction unit 612 sets the length in the sub-scanning direction of intersecting the scanned image 603a in which the change amount of the pixel value is calculated in the main scanning direction as the first size (for example, 30 lines). The first feature amount of the streak-like defect generated in the image formed on the paper Sh is extracted from the classification result 603d for each of the first streak detection regions. At the same time, the defect feature amount extraction unit 612 sets the length in the sub-scanning direction to the second size (for example, 10 lines) shorter than the first size for each of the second streak detection regions of streak-like defects. The second feature amount is extracted from the classification result 603d. At this time, the defect feature amount extraction unit 612 extracts the first feature amount for each first streak detection region from the first average value obtained by averaging the classification values in the sub-scanning direction of the classification result 603d, and also From the second average value obtained by averaging the classification values in the sub-scanning direction of the classification result 603d, the second feature amount is extracted for each second streak detection region. The details of the streak feature amount will be described later with reference to FIGS. 6, 9 and 10.

The quality determination unit 613 compares the first feature amount with the preset defect detection threshold th1 (an example of the first defect detection threshold), and sets the second feature amount to a defect higher than the defect detection threshold th1. Compared with the detection threshold value th2 (an example of the second defect detection threshold value), streak-like defects having different lengths in the sub-scanning direction are detected, and the quality of the image formed on the paper Sh is determined. At this time, the quality determination unit 613 is at a position specified in the main scanning direction in which the first feature amount exceeding the defect detection threshold th1 exists, or in the main scanning direction in which the second feature amount exceeding the defect detection threshold th2 exists. It is determined that there is a streak-like defect at the specified position. The details of the process in which the quality determination unit 613 compares each streak feature amount with the defect detection threshold value to detect streak-like defects are shown in FIGS. 8 to 10 described later.

If the quality determination unit 613 does not detect a streak from the streak feature amount, the quality determination unit 613 determines that the read image 603a is normal. On the other hand, if the quality determination unit 613 detects a streak from the streak feature amount, the quality determination unit 613 determines that the read image 603a is abnormal. Then, the quality determination unit 613 stores the streak detection result 631 including the page number of the page that is the source of the read image 603a in the storage unit 63.

The streak detection result 631 is not only stored in the storage unit 63, but also sent to an external storage device 4 connected to the image inspection device 3. The storage device 4 may be, for example, a USB (Universal Serial Bus) memory, an SSD (Solid State Drive), an HDD (Hard Disk Drive), or the like connected to the image inspection device 3. By sending the streak detection result 631 to the storage device 4, the operator can display the streak detection result 631 stored in the storage device 4 and confirm the contents. The streak detection result 631 may be transferred and saved to a cloud server (not shown) or PC6 connected via the communication I / F unit 61.

Further, the control unit 60 transmits the streak detection result 631 read from the storage unit 63 as necessary to the image forming apparatus 2 or the PC 6 via the communication I / F unit 61. The image forming apparatus 2 can display the streak detection result 631 on the operation display unit 13. Therefore, the operators of the image forming apparatus 2 and the image inspection apparatus 3 can confirm the contents of the streak detection result 631 from the operation display unit 13. The operator can also confirm the content of the streak detection result 631 from the PC 6.

The control unit 60 selects a paper ejection tray (an example of a paper ejection destination) of the paper Sh to be conveyed along the conveying path 41 according to the streak detection result 631. For example, the control unit 60 operates the switching unit 44 to eject the normal paper conveyed to the transport path 42 to the output tray 47, and eject the abnormal paper conveyed to the transport path 43 to the output tray 48. Make it paper.

Further, the control unit 60 performs a page reprinting process (referred to as “recovery processing”) corresponding to the scanned image 603a written in the streak detection result 631 as having streaks through the communication I / F unit 61 to perform the image forming apparatus 2 Can be instructed to. The recovery process is performed automatically by the image inspection system 1 or manually by the operator after the rotating body that has caused the streaks has been cleaned or replaced by the operator.

Next, the procedure for extracting the streak feature amount from the scanned image will be described with reference to FIGS. 5 and 6.
FIG. 5 is a diagram showing an example of a scanned image 603a including a streak 70 to be detected and a classification result 603d.

The enlarged view (1) of the scanned image 603a in the upper part of FIG. 5 shows that a streak 70 to be detected is generated in the center of the scanned image 603a and in the sub-scanning direction. As shown in FIG. 1, the data obtained by averaging the pixel values in the sub-scanning direction has a large fluctuation cycle of density unevenness, so that the streaks 70 may not be detected reliably.

Therefore, the change amount calculation unit 611 selects the pixel of interest 71, selects a pixel near the pixel of interest 71 as the comparison pixel 72, and takes the difference between the pixel of interest 71 and the comparison pixel 72. At this time, the change amount calculation unit 611 takes the difference between the comparison pixel 72 and the pixel value in the direction orthogonal to the streak 70 to be detected with respect to the read image 603a. In this example, the difference between the comparison pixel 72 in the main scanning direction and the pixel value is taken with respect to the pixel of interest 71. The closer the pixels to be compared, the less affected by the density unevenness, but if the pixels to be compared are too close, the influence of streaks cannot be detected. Therefore, the change amount calculation unit 611 compares with pixels separated by half or more of the width of the streak 70 to be detected. Then, the change amount calculation unit 611 generates a difference image 603c1 in which the difference value is calculated at the position of the pixel of interest 71.

Next, the defect feature amount extraction unit 614 extracts the feature amount of the streaks based on the difference value of each pixel included in the difference image 603c1 generated by the change amount calculation unit 611. Here, when the defect feature amount extraction unit 614 takes a difference value in a direction orthogonal to the streak (for example, the main scanning direction), the comparison pixel 72 comes to the streak position and the attention pixel 71 is the streak. A large value whose sign is opposite to that when the position is reached can be obtained (see the graph (3) of FIG. 6 described later).

However, although the value of the defect feature amount obtained from the difference value is larger than that in the place where there is no streak, the value itself is small. Therefore, the change amount calculation unit 611 classifies the values into three values according to whether the difference value is larger than the first classification threshold value A or smaller than the second classification threshold value B, and obtains the classification result 603d. The first classification threshold value A is a value larger than the second classification threshold value B. In the following description, when the first classification threshold value A and the second classification threshold value B are not distinguished, they are collectively referred to as “classification threshold value”.

The enlarged view (2) of the classification result 603d in the lower part of FIG. 5 shows an example in which the enlarged view (1) of the scanned image 603a in the upper part of FIG. 5 is quantified. The classification result 603d is obtained by the change amount calculation unit 611 ternating the difference value of each pixel obtained from the read image 603a. The classification result 603d is represented as having each pixel replaced with a value of any one of "0", "128", and "255". Here, for the sake of simplicity, an example of the classification result 603d colored for each pixel is shown. In order to obtain such a classification result 603d, appropriate multi-value increase is required. Therefore, in the present embodiment, the classification threshold value referred to in the multi-value processing is changed according to the region of the scanned image 603a.

FIG. 6 is a diagram showing how the streak feature amount is extracted from the classification result 603d.
The graph (1) in the upper part of FIG. 6 shows an example of a ternary average value. The defect feature amount extraction unit 614 calculates a quantified average value obtained by averaging the classification result 603d in the streak direction for detection. In this example, it is assumed that the streak direction to be detected by the operator is the sub-scanning direction. Then, the ternary average value is represented as a graph in which a peak is formed when the main scanning position is “15” and a valley is formed when the main scanning position is “18”.

The graph (2) in the middle of FIG. 6 also shows an example of the ternary average value. However, the ternary average value shown in the graph (2) is shifted to the right by 3 pixels (the number of pixels from the attention pixel 71 to the comparison pixel 72) as compared with the graph (1), that is, in FIG. It represents a ternary average value in a state where the indicated pixel 71 is shifted to the position of the comparison pixel 72.

The graph (3) at the bottom of FIG. 6 shows the absolute value of the value obtained by subtracting the graph (2) from the graph (1) as the streak feature amount. From the graph (3), it is shown that the values with opposite signs (for example, the main scanning position is “18”) are calculated to be larger than the values at other locations. Then, the quality determination unit 613 determines that a streak has occurred at this position if the streak feature amount calculated by taking the difference between the values of the inverse signs is larger than the defect detection threshold value C (for example, “150”). To do.

By the way, the lengths of the streaks included in the scanned image 603a vary. In addition, noise is often superimposed on the scanned image 603a. The streaks that are darker (larger gradation difference) than the surrounding pixel values and are shorter than the surrounding pixel values tend to have a large influence of noise, although the continuity of the streak feature amount extracted after the ternation averaging is inconspicuous. On the other hand, thin and long streaks as thin as the surrounding pixel values (small gradation difference) tend to be less affected by noise because the continuity of the streak feature amount extracted after the ternation averaging is conspicuous. Further, even if a long streak can be detected, a short streak may be indistinguishable from noise. Therefore, erroneous detection in which noise is detected as a short streak is likely to occur. A method of detecting streaks having different lengths in this way will be described with reference to FIGS. 7 and later.

<Low resolution processing for streaks>
Next, in order to detect the short streaks and the long streaks included in the scanned image 603a, the process of lowering the resolution of the classification result 603d will be described with reference to FIGS. 7 and 8.
FIG. 7 is a diagram showing how the change amount calculation unit 611 calculates the ternary average value by lowering the resolution of the classification result 603d in the sub-scanning direction.

Here, the change amount calculation unit 611 performs a process of averaging the pixel values of a plurality of pixels in the sub-scanning direction for each predetermined number of lines extending in the main scanning direction. For example, it is assumed that the sub-scanning direction of the classification result 603d is composed of X lines.

On the left side of FIG. 7, an example of calculating the ternary average value of short streaks is shown. As shown in the upper left of FIG. 7, in order to enable the quality judgment unit 613 to detect short streaks, the change amount calculation unit 611 divides the classification result 603d by setting the averaging unit for every 10 lines (the range indicated by the diagonal line). The values of the ternary pixels are averaged for each streak detection area. As a result, as shown in the lower left of FIG. 7, a ternary average image in which the length of the classification result 603d in the sub-scanning direction is compressed to the X / 10 line is generated. Further, in one line of the ternary average image, an "A average" obtained by averaging 10 lines of ternary pixels (A) is stored. Similarly, the resolution reduction processing is performed on the other lines of the classification result 603d. In the present embodiment, the process of averaging the values of the ternary pixels in the sub-scanning direction is called "low resolution processing".

On the right side of FIG. 7, an example of calculating the ternary average value of a long streak is shown. As shown in the upper right of FIG. 7, in order to enable the quality judgment unit 613 to detect long streaks, the change amount calculation unit 611 divides the classification result 603d by setting the averaging unit into 30 lines (the range indicated by the diagonal line). The values of the ternary pixels are averaged for each streak detection area. As a result, as shown in the lower right of FIG. 7, a ternary average image in which the length of the classification result 603d in the sub-scanning direction is compressed to the X / 30 line is generated. Further, in one line of the ternary average image, "A, B, C average" in which 30 lines of ternary pixels (A, B, C) are averaged is stored. Similarly, the resolution reduction processing is performed on the other lines of the classification result 603d.

Using the averaging result for each streak detection region calculated by the resolution reduction process shown in FIG. 7, the defect feature amount extraction unit 612 performs a process of extracting the streak feature amount. An example of the classification result 603d from which the streak feature amount is extracted will be described.

FIG. 8 is a diagram showing an example of the positions of the ternary pixels included in the classification result 603d, the noise 80 appearing in the classification result 603d, the short streaks 82, and the long streaks 81. FIG. 8 shows an example of the classification result 603d before the averaging.

The length of the classification result 603d in the sub-scanning direction is 60 lines. On the left side of the classification result 603d, a long streak 81 having a length of 60 lines in the sub-scanning direction appears. Further, a noise 80 having a length of 10 lines in the sub-scanning direction appears in the upper right side of the classification result 603d, and a short streak 82 having a length of 20 lines in the sub-scanning direction appears in the lower part. In the noise 80, the pixels of "255" represented by the ternary value appear at some intervals as compared with the short streaks 82.

In this embodiment, two types of streak detection regions 85 and 86, which are shown as averaging units, are provided. The size of the streak detection region 85 is defined with the length of the classification result 603d in the sub-scanning direction as the first size (for example, 30 lines). The size of the streak detection region 86 is defined by setting the length of the classification result 603d in the sub-scanning direction as a second size (for example, 10 lines) shorter than the first size. In the present embodiment, the first size is set to be three times or more the second size. However, the length of the first size in the sub-scanning direction may be twice or more the length of the second size in the sub-scanning direction. In this embodiment, both the first and second sizes are defined by the length in the sub-scanning direction. Hereinafter, an example of a process of averaging the ternary values for each of the streak detection regions 85 and 86 and extracting the streak feature amount will be described with reference to FIGS. 9 and 10.

<Extraction of long streak streak features>
FIG. 9 is a diagram showing how the defect feature amount extraction unit 612 extracts the streak feature amount of the long streak 81 from the ternary average value of the long streak 81.

As shown in FIG. 7 described above, when the defect feature amount extraction unit 612 extracts the feature amount of the long streaks 81 from the classification result 603d, it performs a resolution reduction process with 30 lines as the averaging unit. Find the ternary average value. An example of the classification result 603d (trivalent average image 90) subjected to the low resolution processing is shown in the upper part of FIG. 9. Since 30 lines are used as the averaging unit, when the defect feature amount extraction unit 612 performs the resolution reduction processing, the length of the classification result 603d in the sub-scanning direction is compressed to 2 lines (= 60/30). The average image 90 is generated. The ternary average image 90 is temporarily stored in the RAM 603. Then, the defect feature amount extraction unit 612 calculates the streak feature amount for the inspection area (1) 91 for 8 pixels in the main scanning direction and 1 pixel in the sub-scanning direction of the ternary average image 90.

In the graph of the ternary average value shown in the central portion of FIG. 9, the defect feature amount extraction unit 612 performs the same processing as the graph (1) in the upper part of FIG. It is drawn corresponding to the valued average value. Then, in the graph of the streak feature amount shown in the lower part of FIG. 9, the defect feature amount extraction unit 612 performs the same processing as the graph (3) in the lower part of FIG. The streak features extracted from each value shown in the graph of the average value are drawn.

In the graph of the streak feature amount, the defect detection threshold value th1 (an example of the first defect detection threshold value) shown by a broken line is set at the position where the streak feature amount is “100”. The defect detection threshold value th1 is a value set in advance in the parameter 603e. Then, from the graph of the streak feature amount shown in the lower part of FIG. 9, the quality determination unit 613 has a long streak at a position where the streak feature amount exceeds the defect detection threshold th1 and the main scanning position is “2” to “4”. Since 81 is detected, it can be determined that the scanned image 603a, which is the source of the classification result 603d, is abnormal. The quality determination unit 613 can include the main scanning position and the sub-scanning position where the long streaks 81 are detected in the streak detection result 631. By performing the low resolution processing on the classification result 603d in this way, the influence of the noise 80 is reduced, so that the quality determination unit 613 does not erroneously detect the noise 80 as a long streak 81.

<Extraction of noise and streak features of short streaks>
FIG. 10 is a diagram showing how the defect feature amount extraction unit 612 extracts the noise 80 and the streak feature amount of the short streak 82 from the ternary average value of the short streak 82.

Since the streak feature amount after ternation averaging is not conspicuous in the short streak 82, the defect detection threshold th1 for the streak feature amount used in the low resolution processing by the 30-line averaging shown in FIG. 9 is maintained. , The quality determination unit 613 erroneously detects the noise 80 as a short streak 82. Therefore, the quality determination unit 613 sets the defect detection threshold value th2 (second defect detection threshold value) with respect to the streak feature amount extracted from the ternary average value that has been subjected to the low resolution processing in order to detect only the short streaks 82. By changing the setting of (one example) to "194", false detection of noise 80 is avoided. That is, the defect detection threshold value th2 is set to a value higher than the defect detection threshold value th1.

(With noise and no short streaks)
As shown in FIG. 7 described above, when the defect feature amount extraction unit 612 extracts the feature amount of the noise 80 from the classification result 603d, it performs a resolution reduction process with 10 lines as the averaging unit 3 Find the valued average value. An example of the classification result 603d (trivalent average image 95) subjected to the low resolution processing is shown in the upper part of FIG. 10. Since 10 lines are used as the averaging unit, when the defect feature amount extraction unit 612 performs the resolution reduction processing, the length of the classification result 603d in the sub-scanning direction is compressed to 6 lines (= 60/10). The average image 95 is generated. The ternary average image 95 is temporarily stored in the RAM 603. Then, the defect feature amount extraction unit 612 targets the inspection area (2) 96 for 8 pixels in the main scanning direction and 1 pixel in the sub-scanning direction on the upper side of the ternary average image 95, and sets the streak feature amount. calculate.

In the graph of the ternary average value shown in the central portion on the left side of FIG. 10, the defect feature amount extraction unit 612 performs the same processing as the graph (1) in the upper part of FIG. It is drawn corresponding to the ternary average value. The graph of the streak feature amount shown in the lower left side of FIG. 10 is shown in the center part of the left side of FIG. 10 by the defect feature amount extraction unit 612 in the same process as the graph (3) in the lower part of FIG. The streak features extracted from each value shown in the graph of the ternary average value are drawn.

In the graph of the streak feature amount shown in the lower left side of FIG. 10, the defect detection threshold value th2 is set at the position where the streak feature amount is “194”. The defect detection threshold value th2 is a value set in advance in the parameter 603e. Then, from the graph of the streak feature amount shown in the lower left side of FIG. 10, the quality judgment unit 613 shows that the streak feature amount does not exceed the defect detection threshold th2. The noise 80 generated at the position of "7" is not erroneously detected as a short streak 82. The noise 80 detected as the streak feature amount does not affect the quality judgment of the read image 603a by the quality judgment unit 613.

(No noise, with short streaks)
As described above, the defect feature amount extraction unit 612 targets the inspection area (3) 97 for 8 pixels in the main scanning direction and 1 pixel in the sub-scanning direction under the ternary average image 95. Calculate the streak feature amount.

The graph of the ternary average value shown in the center on the right side of FIG. 10 is obtained by the defect feature amount extraction unit 612 in the inspection area (3) 97 by the same processing as the graph (1) in the upper part of FIG. 6 described above. It was drawn correspondingly. Then, the graph of the streak feature amount shown in the lower right side of FIG. 10 is shown in the center part of the right side of FIG. 10 by the defect feature amount extraction unit 612 in the same process as the graph (3) in the lower part of FIG. The streak features are extracted from each value shown in the graph of the ternary average value and drawn.

In the graph of the streak feature amount shown in the lower right side of FIG. 10, the defect detection threshold th2 is set at the position where the streak feature amount is “194”. Then, the quality determination unit 613 detects the short streak 82 at the position where the main scanning position where the streak feature amount exceeds the defect detection threshold th2 is “7” from the graph of the streak feature amount shown in the lower right side of FIG. The scanned image 603a, which is the source of the classification result 603d, can be determined to be abnormal. Then, the quality determination unit 613 can include the main scanning position and the sub-scanning position where the short streaks 82 are detected in the streak detection result 631.

Next, an example of the processing performed by the inspection processing apparatus 5 will be described with reference to FIG.
FIG. 11 is a flowchart showing an example of processing performed by the inspection processing apparatus 5 according to the first embodiment.

First, the change amount calculation unit 611 shifts the read image 603a (for example, FIG. 5) read from the RAM 603 and inputs the image by the number of first pixels (for example, 3 pixels) (S1). Then, the change amount calculation unit 611 generates a difference image 603c1 (for example, FIG. 4) based on the shift image obtained in step S1 and the read image 603a that has not been processed in step S1 (S2).

Next, the change amount calculation unit 611 performs ternation processing of the difference image 603c1 based on the first classification threshold value and the second classification threshold value read from the parameter 603e, and classifies each pixel including the ternary value. The result 603d (eg, FIG. 8) is generated (S3).

Next, the defect feature amount extraction unit 612 extracts the streak feature amount (for example, the lower part of FIGS. 9 and 10) based on the classification result 603d (S4). At this time, as shown in FIGS. 9 and 10, the defect feature amount extraction unit 612 identifies the inspection area specified from the quantified average image generated by low-resolution processing of the classification result 603d with 10 lines and 30 lines. The streak feature amount is extracted for each inspection area.

Next, the quality determination unit 613 compares the extracted streak feature amount with the defect detection threshold value th1 or th2 read from the parameter 603e, detects the streak (S5), and determines the quality of the read image 603a. The streak detection result 631 is written in the storage unit 63. After that, the inspection processing device 5 ends the main processing.

In the inspection processing apparatus 5 according to the first embodiment described above, the difference image 603c1 is obtained by taking the difference between the shift image obtained by shifting the read image 603a by several pixels and the read image 603a, and then obtaining the difference image 603c1. The ternary average value of the classification result 603d obtained by ternating the difference image 603c1 is obtained. At this time, the inspection processing device 5 stores the ternary average value calculated by performing the low resolution processing in the sub-scanning direction of the classification result 603d with different numbers of lines according to the length of the streaks to be detected. The streak feature amount is extracted from the inspection area of the ternary average image. After that, the inspection processing device 5 compares the extracted streak feature amount with the defect detection threshold value acquired from the parameter 603e according to the length of the streak to be detected, and detects the presence or absence of the streak in the read image 603a. .. The inspection processing device 5 can reliably detect each streak even if the scanned image 603a contains streaks having different lengths in the sub-scanning direction. Further, the inspection processing device 5 can detect the streaks even if the region where the streaks are generated in the scanned image 603a is flat, and the streak detection accuracy is improved.

Further, as shown in FIG. 8, the streak detection region 85 corresponding to the long streak 81 averages the ternary values for 30 lines, so that the influence of noise is reduced, so that the defect detection threshold th1 is lowered. You may. Further, even if the defect detection threshold value th1 is lowered, the quality determination unit 613 can reliably detect the thin and long streaks 81. On the other hand, since the streak detection region 86 corresponding to the short streak 82 averages the ternary values for 10 lines, the influence of the noise 80 becomes large. However, since the short streaks 82 tend to appear dark, the noise 80 is not erroneously detected as the short streaks 82 by setting the defect detection threshold value th2 to be higher than the defect detection threshold value th1. Further, the quality determination unit 613 can reliably detect a short streak 82 whose streak feature amount is larger than the defect detection threshold value th2.

It should be noted that the wide streaks have a gentle effect (change in pixel value) due to the streaks. Therefore, if the positions of the pixels to be extracted as the comparison pixels are too close to the pixels of interest, the difference from the comparison pixels becomes small and the streaks cannot be detected. On the other hand, if the comparison pixel is too far from the pixel of interest, the streaks generated in the narrow flat region cannot be detected.

Therefore, the number of first pixels calculated by the change amount calculation unit 611 in order to obtain the difference image 603c1 may be a plurality of different values (for example, 3 pixels, 5 pixels, 7 pixels, 15 pixels). Then, the change amount calculation unit 611 may generate a plurality of difference images 603c1 in which the difference values calculated for each of a plurality of different values are stored at the positions of the pixels of interest. By setting a plurality of shifts of the comparison pixel shifted by the change amount calculation unit 611 with respect to the attention pixel in the parameter 603e and obtaining the difference between the shift amount of one attention pixel and the comparison pixel shifted by a different shift number. , The presence or absence of streaks may be analyzed in parallel. By such a process, the defect feature amount extraction unit 612 may extract the streak feature amount from each of the plurality of difference images 603c1, and the quality determination unit 613 may detect the streak feature amount from the extracted streak feature amount. Thereby, the streaks appearing in the scanned image 603a can be detected regardless of the thickness of the streaks.

[Modified example of the first embodiment]
By the way, when the defect feature amount extraction unit 612 changes the size of the averaging region, streaks straddling the adjacent streak detection region (see FIG. 7) may occur in the resolution reduction processing. When a streak straddles an adjacent streak detection area, a particularly short streak is divided by the streak detection area and is treated as noise. Then, an erroneous detection occurs in which the quality determination unit 613 cannot detect a short streak. In the above-described embodiment, the ternary average value is calculated so that the adjacent streak detection regions defined for every 10 lines or 30 lines do not overlap in the sub-scanning direction. The feature amount extraction unit 612 performs a process of overlapping a part of the adjacent streak detection regions.

In order to enable the quality determination unit 613 to detect short streaks and long streaks contained in the scanned image 603a, the defect feature amount extraction unit 612 lowers the resolution of the classification result 603d and generates a ternary average image. The processing will be described with reference to FIG.
FIG. 12 is a diagram showing an example in which a part of a plurality of streak detection regions overlaps.

Here, too, the defect feature amount extraction unit 612 performs a process of averaging the pixel values of the plurality of pixels in the sub-scanning direction for each predetermined number of lines extending in the main scanning direction. For example, it is assumed that the sub-scanning direction of the classification result 603d is composed of X lines. Then, the defect feature amount extraction unit 612 superimposes a part of the first streak detection region on the other first streak detection region, and for each of the first streak detection region and the other first streak detection region. While extracting the first feature amount and superimposing a part of the second streak detection area on the other second streak detection area, each of the second streak detection area and the other second streak detection area. The second feature amount is extracted.

For example, on the left side of FIG. 12, an example of calculating the ternary average value of short streaks is shown. As shown in the upper left of FIG. 12, in order to enable the quality judgment unit 613 to detect short streaks, the defect feature amount extraction unit 612 classifies the averaging unit from the area A in units of 10 lines (the range indicated by the diagonal line). As a result, the values of the quantified pixels are averaged for each of the streak detection regions 101 divided by 603d. Similarly, the defect feature amount extraction unit 612 averages the values of the ternary pixels for each of the streak detection areas 102 in which the classification result 603d is divided, with the averaging unit set every 10 lines from the area B (range indicated by diagonal lines). To do. As shown in FIG. 12, the streak detection areas 101 and 102 overlap the five lines of the area B. Then, as shown in the lower left of FIG. 12, the sub-scanning direction of the classification result 603d is compressed to the X / 10 line. "A, B average" obtained by averaging 10 lines of ternary pixels (A) is stored in one line of the ternary average image, and "B, C average" is stored in the next one line. Is stored. Similarly, the resolution reduction processing is performed on the other lines of the classification result 603d.

On the right side of FIG. 12, an example of calculating the ternary average value of a long streak is shown. As shown in the upper right of FIG. 12, in order to enable the quality judgment unit 613 to detect long streaks, the defect feature amount extraction unit 612 sets the averaging unit to every 30 lines (range indicated by diagonal lines) and sets the classification result 603d. The values of the ternary pixels are averaged for each of the divided streak detection areas 103. Similarly, the defect feature amount extraction unit 612 averages the values of the ternary pixels for each of the streak detection regions 104 in which the classification result 603d is divided, with the averaging unit set every 30 lines from the region C (range indicated by diagonal lines). To do. As shown in FIG. 12, 20 lines overlap in the streak detection areas 103 and 104 in the range of areas C to F. Then, as shown in the lower right of FIG. 12, the sub-scanning direction of the classification result 603d is compressed to the X / 30 line. One line of the ternary average image stores "A to F average" in which 30 lines of ternary pixels (A to F) are averaged, and the next one line stores 30 lines. The "C to H average" obtained by averaging the ternary pixels (C to H) is stored. Similarly, the resolution reduction processing is performed on the other lines of the classification result 603d.

In this way, the defect feature amount extraction unit 612 extracts the streak feature amount by the processes shown in FIGS. 9 and 10 using the quantified average value averaged for each streak detection region. After that, the quality determination unit 613 can compare the defect detection threshold value with the streak feature amount and detect short streaks or long streaks separated from the noise. Therefore, the quality determination unit 613 determines the pixel value by the defect feature amount extraction unit 612 so that a part of the plurality of streak detection areas overlaps even if the streaks straddle a plurality of adjacent streak detection areas. Since the averaging process is performed, the quality determination unit 613 can detect one streak or a streak detection region range that is easy to detect.

[Second Embodiment]
<Detection of streaks appearing in areas other than the flat image area>
In the above-described embodiment, an example of a process for detecting a streak appearing in a previously known flat image region (referred to as a “flat region”) in the scanned image 603a has been described. If it is a flat region, the classification threshold value may be changed according to the density of the streaks to be detected (gradation difference of pixel values). Therefore, the change amount calculation unit 611 extracts only the flat region from the read image 603a to generate the difference image 603c1, the defect feature amount extraction unit 612 extracts the streak feature amount, and the quality determination unit 613 is extracted. It is sufficient to detect the streaks only in the area.

However, in an actual printed matter, not only a flat region but also a non-flat image region (referred to as "non-flat region") appears. For example, a streak on a non-flat region where there is a gradation in which the amount of fluctuation of the pixel value is small becomes conspicuous. Therefore, in the second embodiment, the process of detecting the streaks appearing in the non-flat region by the inspection processing device 5A will be described.

FIG. 13 is a block diagram showing a configuration example of the control system of the image inspection device 3. In the inspection processing device 5A included in the image inspection device 3 according to the present embodiment, when the RIP image used for forming an image on the paper Sh is sent from the image forming device 2 to the inspection device 3A, the printed matter is printed from the RIP image. The fluctuation amount of the pixel value in the non-flat region of is extracted, and the threshold value for ternation is changed based on the extracted fluctuation amount.

The control unit 60 of the inspection processing device 5A has an image conversion unit 615 and an alignment unit 616 in addition to the change amount calculation unit 611, the defect feature amount extraction unit 612, and the quality judgment unit 613 according to the first embodiment described above. To be equipped. Also in this embodiment, the functions of the image conversion unit 615 and the alignment unit 616 are realized by executing the program read from the ROM 602 by the CPU 601.

The output target image 603b stored in the RAM 603 is a bitmap format image that has been previously subjected to rasterization processing (RIP processing) by the control unit 50 of the image forming apparatus 2. The output target image 603b is the source of the image formed on the paper Sh by the image forming apparatus 2. An image determined in advance by the operator to be correct and read in advance by the reading unit 45 may be used as the output target image 603b.

Since the output target image 603b is used by the image forming apparatus 2 to form an image on the paper Sh, the CMYK color mode is set. Therefore, the image conversion unit 615 performs image conversion that matches the color mode of the output target image 603b with the color mode of the scanned image 603a. Here, the color mode of the scanned image 603a is RGB. Therefore, the image conversion unit 615 converts the color mode of the output target image 603b from CMYK to RGB. In the following description, the output target image 603b will be described assuming that the color mode has been converted to RGB.

The alignment unit 616 aligns the print position of the output target image 603b with the scanned image 603a read from the paper Sh on which the image is formed (referred to as "alignment"). At this time, the alignment unit 616 aligns the read image 603a read from the RAM 603 with the position of the output target image 603b stored in advance in the RAM 603.

Then, the change amount calculation unit 611 generates the difference image 603c2 by taking the difference between the read image 603a aligned by the alignment unit 616 and the output target image 603b. The difference image 603c2 is stored in the RAM 603 together with the difference image 603c1 obtained by taking the difference between the shift image of the read image 603a and the read image 603a before the shift described in the first embodiment.

Further, the change amount calculation unit 611 generates a threshold image 603f to be referred to for creating the classification result 603d based on the difference image 603c2. At this time, the change amount calculation unit 611 changes the classification threshold value according to the change in the pixel value of the read image 603a aligned by the alignment unit 616. The threshold image 603f is two types of images defined by the first classification threshold A and the second classification threshold B shown in FIG. 14, which will be described later, and is stored in the RAM 603. The change amount calculation unit 611 digitizes the difference image 603c1 based on the generated threshold image 603f, and obtains the classification result 603d.

FIG. 14 is a graph showing an example of pixel values, difference values, and classification thresholds of pixels in the flat region and the non-flat region of the scanned image 603a. The horizontal axis of this graph represents the main scanning position, and the vertical axis represents three types of values (pixel value, difference value, and classification threshold value). Further, the flat region and the non-flat region are defined by the graph of the difference value on the upper side of FIG.

The graph on the lower side of FIG. 14 shows the fluctuation of the pixel value of the output target image 603b subjected to the RIP processing. Further, in the graph on the upper side of FIG. 14, in the graph on the lower side of FIG. 14, the pixel of interest and the pixel of interest are shifted by 3 pixels in the horizontal direction (direction orthogonal to the streak to be detected) in the same manner as in FIG. It represents the variation of the difference value of the difference image 603c2 obtained by taking the difference from the comparison pixel. In the figure, the first classification threshold value A and the second classification threshold value B are described as “threshold value A” and “threshold value B”, respectively.

The pixel value of the output target image 603b is represented by a graph of a polygonal line L1. From the graph of the polygonal line L1, it is shown that the pixel value and the difference value do not change because the flat region is between “1” and “7” of the main scanning position. On the other hand, since the area from "7" to "33" of the main scanning position is a non-flat region, the pixel value and the difference value change. For example, the pixel value in the non-flat region gradually increases from "11" to "21" of the main scanning position and keeps a constant value, and then gradually decreases from "23" to "33" to the original value. Return to "128".

For example, the difference value is represented by a graph of a polygonal line L10. As shown in the graph of the polygonal line L10, the difference value in the non-flat region gradually decreases from "7" to "9" at the main scanning position, takes a constant negative value from "9" to "17", and becomes " It gradually increases from 17 "to" 23 ". Then, it takes a constant positive value from "23" to "30" of the main scanning position, gradually decreases from "30" to "33", and returns to the original value "0".

Here, a value obtained by adding a predetermined value (for example, "8") to the difference value is called a first classification threshold value A, and a value obtained by subtracting a predetermined value (for example, "8") from the difference value is called. , Called the second classification threshold B. The first classification threshold value A is represented by the graph of the polygonal line L11, and the second classification threshold value B is represented by the graph of the polygonal line L12.

In the following description, when the first classification threshold value A and the second classification threshold value B are not distinguished, they are collectively referred to as “classification threshold value”. The classification threshold is stored in parameter 603e. The classification thresholds are all referred to by the change amount calculation unit 611 for ternating the pixels specified by the main scanning position of the scanned image 603a.

FIG. 15 is a flowchart showing an example of processing performed by the inspection processing device 5A according to the second embodiment. In this process, the processes of steps S11 and S12 are the same as the processes of steps S1 and S2 of FIG. Therefore, after explaining the processing after step S21, the processing after step S13 will be described.

First, the image conversion unit 615 converts the color mode of the output target image 603b read from the RAM 603 from CMYK to RGB (S21). By this process, the image conversion unit 615 matches the color mode of the output target image 603b (RIP image) received from the image forming device 2 by the image inspection device 3 with the color mode of the scanned image 603a.

Next, the alignment unit 616 performs alignment processing on the output target image 603b converted to RGB so that the position is aligned with the scanned image 603a (S22). Next, the change amount calculation unit 611 shifts the output target image 603b by the number of first pixels (S23). In the present embodiment, the image of the output target image 603b is shifted in the horizontal direction by three pixels.

Next, the change amount calculation unit 611 generates a difference image 603c2 based on the shift image obtained in step S23 and the output target image 603b that has not been processed in step S23 (S24).

Next, the change amount calculation unit 611 sets a value obtained by adding a predetermined value for each pixel value of each pixel included in the difference image 603c2 as the first classification threshold value A, and sets a value obtained by subtracting the predetermined value as the second classification threshold value B. The threshold image 603f is generated (S25). The threshold image 603f is temporarily stored in the RAM 603.

After steps S12 and S25, the change amount calculation unit 611 performs the quantification process of the difference image 603c1 based on the threshold image 603f read from the RAM 603, and generates the classification result 603d (S13). Next, the defect feature amount extraction unit 612 calculates the quantified average value for each streak detection region obtained by dividing the classification result 603d by 10 lines or 30 lines in the sub-scanning direction, and then the streak feature amount (FIGS. 9 and 9). (See the lower part of 10) is extracted (S14).

Then, the quality determination unit 613 compares the streak feature amount with the defect detection threshold value read from the parameter 603e, and detects the streak (S15). The quality determination unit 613 writes the streak detection result 631 in the storage unit 63. After that, the inspection processing device 5A ends the main processing.

The inspection processing device 5A according to the second embodiment described above generates a threshold image 603f defined by a classification threshold value with respect to the difference value. Then, the ternation process of the difference image 603c1 is performed based on the threshold image 603f. Therefore, even for the streaks generated in the non-flat region of the scanned image 603a, the pixel values are quantified by the ternary processing and the streak feature amount is extracted, so that the streak detection accuracy can be improved.

[Modification example]
In each of the above-described embodiments, the image inspection device 3 is combined with the inspection processing devices 5 and 5A, but the functions of the inspection processing devices 5 and 5A are incorporated into, for example, the PC 6 to incorporate the inspection processing devices 5 and 5A. It may be separated from the image inspection device 3. Further, the image forming apparatus 2 may have the functions of the inspection processing devices 5 and 5A, and the image forming apparatus 2 may perform an image inspection by itself. Further, by providing a server having the functions of the inspection processing devices 5 and 5A to configure the image forming system, even if the server accumulates the scanned image 603a and the output target image 603b read from the paper Sh by the image inspection device 3. Good. Then, the server may communicate with the image inspection device 3 to detect the streaks of the read image 603a received from the image inspection device 3 and transmit the streak detection result to the image inspection device 3 or the PC 6.

It should be noted that the present invention is not limited to the above-described embodiment, and it goes without saying that various other application examples and modifications can be taken as long as the gist of the present invention described in the claims is not deviated.
For example, the above-described embodiment describes in detail and concretely the configurations of the apparatus and the system in order to explain the present invention in an easy-to-understand manner, and is not necessarily limited to those including all the described configurations. Further, it is possible to replace a part of the configuration of the embodiment described here with the configuration of another embodiment, and further, it is possible to add the configuration of another embodiment to the configuration of one embodiment. It is possible. It is also possible to add, delete, or replace a part of the configuration of each embodiment with another configuration.
In addition, the control lines and information lines indicate those that are considered necessary for explanation, and do not necessarily indicate all the control lines and information lines in the product. In practice, it can be considered that almost all configurations are interconnected.

1 ... Image inspection system, 2 ... Image forming device, 3 ... Image inspection device, 5 ... Inspection processing device, 30 ... Image forming unit, 45 ... Reading unit, 60 ... Control unit, 603a ... Scanned image, 603b ... Output target image , 603c1, 603c2 ... Difference image, 603d ... Classification result, 603e ... Parameter, 603f ... Threshold image, 611 ... Change amount calculation unit, 612 ... Defect feature amount extraction unit, 613 ... Quality judgment unit 613 ... Image conversion unit, 616 ... Alignment part, 631 ... Streak detection result

Claims (8)

Of the plurality of pixels included in the scanned image read from the recording material on which the image is formed, the comparative pixels that are separated by the first pixel number in the first direction with respect to the pixel value of the pixel of interest with respect to the pixel value of interest. A change amount calculation unit that calculates the difference value of the pixel value and classifies the difference value into a plurality of classification values by the classification threshold as the change amount of the pixel value.
A streak-like defect generated in the image formed on the recording material for each first defect detection region having the length of the second direction intersecting the first direction of the classification result as the first size. The first feature amount of the above is extracted from the classification result, and the length in the second direction is set to the second size shorter than the first size. A feature amount extraction unit that extracts the second feature amount of the defect from the classification result, and a feature amount extraction unit.
The second feature amount is compared with the first defect detection threshold value, and the second feature amount is compared with the second defect detection threshold value higher than the first defect detection threshold value. An image inspection apparatus including a quality determination unit that detects the streak-like defects having different lengths in directions and determines the quality of the image formed on the recording material. The feature amount extraction unit extracts the first feature amount for each of the first defect detection regions from the first average value obtained by averaging the classification values in the second direction, and the second feature amount extraction unit. The image inspection apparatus according to claim 1, wherein the second feature amount is extracted for each of the second defect detection regions from the second average value obtained by averaging the classification values in the direction. The quality determination unit is at a position specified in the first direction in which the first feature amount that exceeds the first defect detection threshold exists, or the second feature amount that exceeds the second defect detection threshold. The image inspection apparatus according to claim 2, wherein it is determined that there is a streak-like defect at a position specified in the first direction in which the image is present. An image conversion unit that converts the color mode of the output target image, which is the source of the image formed by the image forming apparatus on the recording material, according to the color mode of the scanned image.
A positioning unit for aligning the position of the output target image converted by the image conversion unit with the read image is provided.
The image inspection apparatus according to claim 3, wherein the change amount calculation unit changes the classification threshold value according to a change in the pixel value of the read image aligned by the alignment unit. The feature amount extraction unit superimposes a part of the first defect detection region on the other first defect detection region, and for each of the first defect detection region and the other first defect detection region. While extracting the first feature amount and superimposing a part of the second defect detection region on the other second defect detection region, the second defect detection region and the other second defect detection region. The image inspection apparatus according to any one of claims 1 to 4, wherein the second feature amount is extracted for each defect detection region. The image inspection apparatus according to any one of claims 1 to 5, wherein the first size is three times or more the second size. The image inspection apparatus according to any one of claims 1 to 6, further comprising a reading unit that reads the image from the recording material and generates the read image. An image forming apparatus for forming an image on a recording material and an image inspection apparatus for inspecting the image formed on the recording material are provided.
The image inspection device is
Of the plurality of pixels included in the scanned image read from the recording material, the difference between the pixel values of the pixels of interest and the pixel values of the comparison pixels separated by the number of first pixels in the first direction with respect to the pixels of interest. A change amount calculation unit that calculates a value and classifies the difference value into a plurality of classification values by a classification threshold as a change amount of the pixel value.
A streak-like defect generated in the image formed on the recording material for each first defect detection region having the length of the second direction intersecting the first direction of the classification result as the first size. The first feature amount of the above is extracted from the classification result, and the length in the second direction is set to the second size shorter than the first size. A feature amount extraction unit that extracts the second feature amount of the defect from the classification result, and a feature amount extraction unit.
The second feature amount is compared with the first defect detection threshold value, and the second feature amount is compared with the second defect detection threshold value higher than the first defect detection threshold value. An image inspection system including a quality determination unit that detects the streak-like defects having different lengths in directions and determines the quality of the image formed on the recording material.

Images