JP2015210652A - Image processor, and image processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable the simplification and the high efficiency achievement of image processing for recognizing divided defects to be the same defect in the case that a defect of an object to be inspected is divided into both ends of image data after polar coordinate conversion due to a starting position of polar coordinate conversion processing.SOLUTION: An image processor includes a storage part which can store image data acquired by imaging the object to be inspected, and a control part for executing polar coordinate conversion processing, binarization processing, and labeling processing. The control part gives the same label value because both pixels are connected in the case that a pixel with a label value to be given thereto respectively exists in one end 120d corresponding to the starting position in a circumferential direction of polar coordinate conversion processing of image data 120 subjected to binarization processing in labeling processing, and in the other end 120e corresponding to the ending position in the circumferential direction.

Description

  The present invention relates to an image processing apparatus and an image processing method for executing a labeling process on acquired image data.

  Conventionally, for example, an optical member such as a lens (hereinafter referred to as “inspected object”) is inspected for the presence of defects (dust, scratches, etc.) using image processing to determine whether the product is defective or not (hereinafter referred to as “defect determination”). Is known). In such an inspection apparatus, an inspection object irradiated with light from a light source is photographed with a camera, and a predetermined image processing is performed on the captured image data, thereby extracting defects and performing an inspection for determining a defect. doing. As the predetermined image processing, for example, the inspection apparatus executes a labeling process in order to acquire a feature amount such as an area or length of an object shown in the image data. Here, the labeling process is a threshold value or higher for image data on which a binarization process for binarizing a value of a pixel (hereinafter referred to as “pixel value”) with a predetermined threshold as a boundary is executed. This is a process of assigning the same label value to a region where pixels having a pixel value of are connected.

  In the inspection apparatus described above, feature amount analysis processing for extracting feature amounts such as the area and length of defects in the image data is performed on the image data on which the labeling processing has been performed. In the feature amount analysis process, the inspection apparatus refers to the image data that has been subjected to the labeling process, and extracts the feature amount of the defect that has been given the same label value by the labeling process. Then, in the defect determination, the inspection apparatus determines that the defect exists in the inspection object and is defective when the feature amount of the defect extracted by the feature amount analysis process is a certain value or more. .

  Here, if the object to be inspected has a point-symmetric shape, such as a paper cup, iron pipe, or lens, the required features include features such as the length in the rotational direction and the distance from the center. An amount may be required. Since these feature amounts are difficult to calculate by feature amount analysis from an orthogonal coordinate system image, polar coordinate expansion processing is often performed and labeling processing is performed on the polar coordinate system image. However, when the polar coordinate conversion process is executed on the image data obtained by imaging the inspection object, if the defect is located across the start coordinate and the end coordinate of the polar coordinate conversion process, one defect is detected. Will be divided at both ends of the polar coordinate transformed image data. In such a case, when the labeling process is executed, separate label values are given to one defect divided at both ends of the image data. Therefore, the inspection apparatus executes the feature amount analysis process in the feature amount analysis process on the assumption that one defect that is divided at both ends of the image data and is given a separate label value is a separate defect. End up. As a result, the inspection apparatus extracts a feature amount smaller than the feature amount of one defect from each of the divided defects, and there arises a problem that the defect determination cannot be performed accurately in the above-described defect determination. Sometimes.

  Patent Documents 1 and 2 are methods for solving the problem of defect defect determination caused by the starting coordinates of polar coordinate conversion processing when performing defect determination using image data subjected to polar coordinate conversion processing. Has been proposed.

  For example, in the method described in Patent Document 1, when a defect is found at one end of image data subjected to polar coordinate conversion processing, a part of one end including the found defect is reversed and copied. Then, a defect determination is executed after connecting a part of one end that has been reversed left and right to the other end, thereby preventing a defect from being problematic.

  Moreover, in the method described in Patent Document 2, by performing the polar coordinate conversion process over 360 ° or more in the polar coordinate conversion process, it is possible to prevent the defect from being divided and to prevent a problem from being caused in the defect defect determination. doing.

JP 2003-240727 A JP 2005-265488 A

  However, in the invention described in Patent Document 1, after extracting the defect once, a part of one end of the image data including the defect is copied, connected to the other end of the image data, and then extracted again. Yes. Therefore, in the invention described in Patent Document 1, the defect determination is executed after extracting the defect a plurality of times, and there is a problem that it takes time until the inspection result is output due to complicated image processing. .

  Further, in the invention described in Patent Document 2, by performing the polar coordinate conversion process over 360 ° or more, the data capacity is larger than the image data that can be acquired when the polar coordinate conversion process is performed over the originally necessary 360 °. It gets bigger. For this reason, in the invention described in Patent Document 2, a large memory area for storing image data is required, and since the image data for executing image processing is large, each image such as feature amount analysis processing is used. There was a problem that processing took time.

  Accordingly, an object of the present invention is to provide an image processing apparatus and an image processing method capable of simplifying image processing and improving efficiency.

  An image processing apparatus according to the present invention includes a storage unit capable of storing image data acquired by imaging an object to be inspected, and a rotation center point of a region to be measured in the image data stored in the storage unit as a starting point. The polar coordinate conversion process is executed from the circumferential start position to the circumferential end position connected to the start position, and pixels constituting the image data on which the polar coordinate conversion process has been executed are set with a predetermined threshold as a boundary. A binarization process for binarizing the first pixel value and the second pixel value is executed, and either the first pixel value or the second pixel value is obtained by the binarization process. A control unit that executes a labeling process that assigns the same label value to each of the connected pixels when pixels converted into pixel values are connected, and the control unit includes the labeling process In the above binarization processing When there is a pixel to which the label value is added at one end corresponding to the start position and the other end corresponding to the end position of the image data that has been executed, the image data exists at the one end and the other end. It is characterized in that the same label value is given as if both pixels are connected.

  Further, in the image processing method according to the present invention, the control unit has the start position from the circumferential start position starting from the rotation center point of the measurement target region in the image data of the inspection object stored in the storage unit. A polar coordinate conversion processing step for executing the polar coordinate conversion processing up to the end position in the circumferential direction connected to the control unit, and the control unit has predetermined pixels constituting the image data on which the polar coordinate conversion processing has been executed in the polar coordinate conversion processing step A binarization processing step for executing binarization processing for binarization into a first pixel value and a second pixel value with a threshold as a boundary; and a control unit that performs the binarization in the binarization processing step. In the image data that has been subjected to the conversion processing, when pixels converted to either the first pixel value or the second pixel value are connected, the connected pixels Labels that give the same label value to each A labeling process step for executing a labeling process, wherein the control unit corresponds to one end corresponding to the start position and the end position of the image data on which the binarization process has been performed in the labeling process step. When there is a pixel to which the label value is provided at each of the other ends, end processing for giving the same label value as the one end and both pixels existing at the other end are connected It is characterized by executing.

  According to the present invention, in the labeling process, the same label value is given assuming that both pixels existing at one end and the other end of the image data are connected. As a result, even if one defect is divided into both ends of the image data according to the start coordinate of the polar coordinate conversion process, the same label value is given as one connected defect. be able to. As a result, for example, when one defect is divided at both ends of the image data, it is not necessary to perform image processing such as correcting the image data by complicated processing, simplifying image processing, and increasing efficiency. Can be achieved.

Schematic which shows an inspection apparatus. 1 is a block diagram illustrating a configuration of an image processing apparatus. The figure which shows the image data before the coordinate transformation acquired with the camera. (A) is a figure which shows the image data after performing a polar coordinate conversion process to the image data acquired with the camera. (B) is a figure which shows the image data after performing a binarization process to the image data after performing a polar coordinate conversion process. It is a figure which shows the pixel used as investigation object among the pixels connected to the attention pixel which concerns on this Embodiment, (a) is a pixel which is neither a start column of an image data, nor an end column, but a attention pixel. The figure which shows the pixel used as the investigation object in a certain case. FIG. 6B is a diagram illustrating a pixel to be investigated when a pixel in the start column of image data is a target pixel. (C) is a figure which shows the pixel used as the investigation object when the pixel of the end row | line of image data is an attention pixel. The flowchart which shows the labeling process which concerns on this Embodiment. The flowchart which shows the label value provision process which concerns on this Embodiment. The flowchart which shows the label value update process which concerns on this Embodiment. It is the schematic which shows the specific example of the labeling process which concerns on this Embodiment, (a) is a figure which shows the image data at the time of a labeling process start. (B) thru | or (d) is a figure which shows the case where an attention pixel is connected with the pixel which has a label value. FIG. 6E is a diagram illustrating a state in which label value assignment processing has been performed on all the pixels constituting the image data. FIG. 6F is a diagram illustrating a state in which label value update processing has been performed on all pixels constituting image data and labeling processing has been completed. (A) is a figure which shows the look-up table which concerns on this Embodiment. (B) is a diagram showing a post-update lookup table.

  Hereinafter, embodiments according to the present invention will be described with reference to FIGS. 1 to 10. The inspection apparatus 1 shown in FIG. 1 acquires image data by imaging the inspection object 11, a holder 12 as a positioning member for positioning the inspection object 11, a light source 13 that irradiates the inspection object 11 with light. And an image processing apparatus 300 as an imaging means.

  The inspection object 11 is, for example, an optical element such as a lens. However, the inspection object 11 is not limited to this as long as the region to be measured in the inspection object 11 has a point-symmetric shape. Note that the shape of the inspection object 11 itself may not be point-symmetric, or the background may be reflected in the image data obtained by imaging. That is, the present invention can be applied if the region to be measured is generally point-symmetric with respect to a certain rotation center. In the following description, the term “point symmetry” is used. This indicates that the rotational symmetry starting from the rotation center in the region to be measured exhibits a degree of symmetry that can be converted into polar coordinates with sufficient accuracy for measurement. It is not necessary to have point symmetry in a strict sense. The holder 12 positions the inspection object 11 so that a point 11b where the measurement target region in the inspection object 11 is substantially point-symmetric is the center of imaging by the camera 14. The light source 13 irradiates light toward the inspection object 11. When light is emitted from the light source 13, when the defect 11a exists in the inspection object 11, the irradiated light is scattered in the defect 11a. The camera 14 includes a camera body 14a and an imaging lens 14b. The camera 14 collects light by the imaging lens 14b and forms an image on a sensor provided in the camera body 14a, thereby acquiring image data. The image data captured by the camera 14 can be transmitted to the image processing apparatus 300. In other words, the image processing apparatus 300 is configured to acquire image data including the inspection object 11. Note that an area sensor, a line sensor, or the like may be used as the imaging device.

  The inspection apparatus 1 irradiates the inspection object 11 with light from the light source 13 and condenses the light scattered at the defect 11a existing on the inspection object 11 by the imaging lens 14b. The inspection apparatus 1 forms an image of the light scattered in the defect 11a collected by the imaging lens 14b with a sensor provided in the camera body 14a, thereby obtaining the image data 110 of the inspection object 11 including the defect 11a. Acquired and transmitted to the image processing apparatus 300.

  Next, the structure of the image processing apparatus 300 will be described with reference to FIG. The image processing apparatus 300 includes a CPU 301 that functions as a control unit, a ROM 302, a RAM 303, an HDD (hard disk drive) 304, a recording disk drive 305, and various interfaces 306 to 308 that function as storable storage units. . Note that the image processing apparatus 300 is a general computer.

  A ROM 302, a RAM 303, an HDD 304, a recording disk drive 305, and various interfaces 306 to 308 are connected to the CPU 301 via a bus 310. For example, the HDD 304 (or the recording disk 331 or the external storage device 322) stores various programs and filter data for operating the CPU 301. Of these various programs, the coordinate conversion program 304a is a program for causing the CPU 301 to perform coordinate conversion of image data, as will be described in detail later. In addition, as will be described in detail later, the image processing program 304b executes each filter processing such as HPF (high pass filter) processing, binarization of image data by binarization processing, and label value addition by labeling processing, as will be described in detail later. It is a program to make it. Each of these programs can be stored in an HDD (hard disk drive) 304, a recording disk drive 305, an external storage device 322, or the like as a recording medium.

  On the other hand, the camera 14 (see FIG. 1) is connected to the interface 306, the monitor 321 is connected to the interface 307, and an external storage device (for example, an external HDD or a non-volatile memory) 322 is connected to the interface 308. ing. From the camera 14, for example, based on a command from the CPU 301, the image data 100 imaged around the point symmetry point 11 b as shown in FIG. 3 is acquired via the interface 306 and the bus 310, and temporarily stored in the RAM 303 or the like. Store in memory. By transmitting output data such as results when image processing described later and defect determination of the inspection object 11 are completed, the monitor 321 displays various images indicating the results.

  In the present embodiment, a description is given of obtaining image data 100 captured by the camera 14 and performing coordinate transformation described later. However, the acquired image data may be stored in advance in the HHD 304, recorded in the recording disk 331, or recorded in the external storage device 322.

  Next, image processing according to the present embodiment will be described. In the image processing method according to the present embodiment, first, the image processing apparatus 300 executes a polar coordinate conversion process for executing a polar coordinate conversion process on image data of the original orthogonal coordinate system. Next, in the image processing method, each pixel constituting the image data subjected to polar coordinate conversion processing is binarized into a first pixel value and a second pixel value with a predetermined threshold as a boundary. The image processing apparatus 300 executes a binarization process that executes the conversion process. Next, in the image processing method, the image processing apparatus 300 executes a labeling process for executing a labeling process for assigning a label value to the pixels constituting the image data that has been subjected to the binarization process.

  In the image processing method, in the labeling process step, it is determined whether or not a pixel to which a label value has been assigned is connected to a pixel to which a label value is to be assigned. And when it determines with connecting, the label value provision process which assign | provides the minimum label value among the label values provided to the pixel which is connected to the pixel which becomes the object which provides a label value is performed. This is executed by the image processing apparatus 300. In the image processing method, in the labeling process step, when the label value assigned to each connected pixel is different, the minimum label value assigned to each pixel is assigned to each pixel. The label value update process to be applied is executed by the image processing apparatus 300. In addition, in the image processing method, when there is a pixel to which a label value is provided at each of one end and the other end of the image data, end processing that provides the same label value as a combination of both pixels Is executed by the image processing apparatus 300. In the image processing method, in the edge processing, the label value addition process and the label value update process are executed by the image processing apparatus 300, so that the same pixel is present at one end and the other end of the image data. The label value of is assigned.

  In the present embodiment, the CPU 301 executes the coordinate conversion program 304 a to execute polar coordinate conversion processing on the image data 100 acquired by the camera 14 and stored in the RAM 303. Specifically, as illustrated in FIG. 3, the CPU 301 uses the orthogonal coordinates (X, Y) of the image data 100 acquired by the camera 14 as a starting point from a point 11 b that is the center point of rotation of the measurement target region. Polar coordinate conversion processing for converting to polar coordinates (r, θ) is executed. In the case of the present embodiment, since the inspection object 11 itself has rotational symmetry, the point 11b is also a point symmetry point 11b included in the inspection object 11 itself. In the present embodiment, as shown in FIG. 3, the CPU 301 is connected to the start position 11c from the start position 11c in the circumferential direction around the point symmetry point 11b of the object 11 to be inspected. The polar coordinate conversion process is executed over 360 ° up to 11d.

  FIG. 4A shows polar coordinate image data 110 that has been subjected to polar coordinate conversion processing by the CPU 301. As shown in FIG. 4A, the image data 100 of the inspection object 11 centered on the point-symmetric point 11b is converted into rectangular polar coordinate image data 110 by the polar coordinate conversion process. Here, the left end of the polar coordinate image data 110 corresponds to the circumferential start position 11c of the polar coordinate conversion process. The right end of the polar coordinate image data 110 corresponds to the circumferential end position 11d of the polar coordinate conversion process.

  Next, the CPU 301 executes the image processing program 304b to perform processing for removing background information and noise such as the reflection 13a of the lighting device 13 from the polar coordinate image data 110 shown in FIG. Here, with respect to the background removal, a process of removing the reflection 13a of the illumination generated in the θ direction by using the HPF in the θ direction and a process of removing the reflection by a mask process or the like are performed. As for noise removal, noise removal is performed by executing LPF (low-pass filter) filter processing, median filter processing, and the like.

  Further, the CPU 301 executes the image processing program 304b to binarize each pixel constituting the polar coordinate image data 110 into a first pixel value and a second pixel value with a predetermined threshold as a boundary. Execute binarization processing. By executing the HPF process and the binarization process, the CPU 301 can obtain the binarized image data 120 in which only the defect shown in FIG. Here, the left end of the binarized image data 120 corresponds to the start position 11c in the circumferential direction of the polar coordinate conversion process, and constitutes one end of the image data in the present embodiment. The right end of the binarized image data 120 corresponds to the end position 11d in the circumferential direction of the polar coordinate conversion process, and constitutes the other end of the image data in the present embodiment. Further, in the present embodiment, the pixel whose pixel value is converted into the first pixel value by the binarization process is the black background pixel in FIG. In the present embodiment, the pixel whose pixel value is converted into the second pixel value by the binarization process is a white background pixel in FIG.

  Here, for example, as shown in FIG. 3, a case where there is a defect 11a over the start position 11c in the circumferential direction and the end position 11d in the circumferential direction of the polar coordinate conversion process will be described. In such a case, when the polar coordinate conversion process is executed, the defect 11a is divided (defects 110a and 110b) at both ends of the polar coordinate image data 110 as shown in FIG. When the labeling process to be described later is performed after the binarization process is performed in a state where the defect 11a is divided into the defect 110a and the defect 110b, the CPU 301 treats the defect 110a and the defect 110b as separate defects and performs labeling. Value. Therefore, a special image processing method is required in order to assign a label value assuming that the defect 110a and the defect 110b are one defect 11a.

  Therefore, in the present embodiment, in the labeling process, when a label value is given to the right edge defect 110b, a label value giving process and a label value updating process, which will be described in detail later, are executed. Thereby, in this Embodiment, the defect 110a and the defect 110b can give the same label value as what is one defect 11a. Details of the image processing will be described below.

  In the present embodiment, the CPU 301 executes a labeling process by a raster scan method from the left end to the right end of the binarized image data 120 by executing the image processing program 304b. Here, in general, the raster scanning method is to first perform labeling processing on each pixel while shifting the pixel by 1 Pixel to the right in the horizontal direction from the pixel at the left end and the upper end of the image data to the pixel at the right end of the image data. Execute the labeling process. Next, the target pixel of the labeling process is shifted by 1 pixel in the vertical direction downward, and shifted again by 1 pixel from the left end to the right in the horizontal direction to execute the labeling process for each pixel, thereby labeling all the pixels constituting the image data. It is a method to execute.

  In the present embodiment, the CPU 301 is configured to execute a labeling process from the left end to the right end of the binarized image data 120, but is not limited thereto. When the CPU 301 is configured to execute the labeling process in one direction, the direction is not limited. In the present embodiment, the CPU 301 is configured to execute a labeling process on each pixel by one pixel. However, it is only necessary that the CPU 301 can execute the labeling process on all pixels constituting the image data. That is, the number of pixels to be subjected to each labeling process is not particularly limited.

  In the execution of the labeling process, the CPU 301 determines whether the pixel that is the target of the labeling process has a first pixel value or a second pixel value. When the CPU 301 determines that the pixel has the second pixel value, the CPU 301 determines that the pixel is a pixel to which a label value is applied, and assigns the label value. In the following description, a pixel to be processed is referred to as a target pixel. In this embodiment, the label value is assigned when the pixel of interest has the second pixel value. However, the label value is assigned when the pixel of interest has the first pixel value. You may comprise.

  Here, in the process of giving the label value (hereinafter referred to as “label value giving process”), the pixel connected to the target pixel that is the target of the determination as to whether or not it has the second pixel value. The assigned label value is investigated, and the label value is assigned to the target pixel. As a survey method for investigating the label value assigned to the pixel connected to the target pixel, it is generally referred to as a so-called four-neighborhood in which the pixel on the left of the pixel of the target pixel and the pixel on the pixel are investigated. Survey methods are known. In addition to the four neighborhoods, a so-called 8-neighbor search method is also known, in which 1 pixel left and top pixels of the pixel of interest and 1 pixel right and top pixels are searched.

  FIG. 5 is a diagram illustrating pixels that the CPU 301 investigates in the labeling process according to the present embodiment. FIG. 5A shows a case where the pixel 301 that is not the pixel at the left end or the pixel at the right end of the binarized image data 120 is the target pixel 120a, and the CPU 301 is connected to the target pixel 120a. It shows the case of investigating the pixel. In the investigation of 4 neighborhoods, the CPU 301 investigates the 0th pixel that is 1 pixel to the left of the pixel of interest 120a and the 1st pixel that is on 1 pixel of the pixel of interest 120a shown in FIG. Further, in the investigation of 8 neighborhoods, in addition to the pixels investigated in the 4 neighborhoods, the CPU 301 adds the second pixel existing 1 pixel left and above the pixel of interest 120a shown in FIG. We will also investigate the 3rd pixel that is 1 pixel right and above.

  Here, in the following description, as shown in FIG. 9A to be described later, the left end of the binarized image data 120, that is, the binarized image data 120 corresponding to the circumferential start position 11c of the polar coordinate conversion processing. The one end 120d is also referred to as a start row. In the following description, the right end of the binarized image data 120, that is, the other end 120e of the binarized image data corresponding to the end position 11d in the circumferential direction of the polar coordinate conversion process is also referred to as an end column.

  FIG. 5B shows a case where the CPU 301 investigates pixels connected to the target pixel 120a when the pixel existing in the start column of the binarized image data 120 is the target pixel 120a. In the investigation of 4 neighborhoods, the CPU 301 investigates the 0th pixel existing on 1 Pixel of the pixel of interest 120a shown in FIG. Further, in the investigation of the vicinity of 8, the CPU 301 also investigates the first pixel located on the right side of the pixel of interest 120a shown in FIG. 5B and the first pixel in addition to the pixels investigated in the vicinity of 4.

  FIG. 5C shows a case where the CPU 301 investigates pixels connected to the target pixel 120a when the pixel existing in the end column of the binarized image data 120 is the target pixel 120a. In the investigation of 4 neighborhoods, the CPU 301 determines that the 0th pixel that is 1 pixel to the left of the target pixel 120a, the 1st pixel that is 1 pixel above the target pixel 120a, and the target pixel 120a illustrated in FIG. The second pixel of the pixel in the start column of the same row is investigated. In the investigation of 8 neighborhoods, the CPU 301 adds 2 pixels present in the start column and the 3rd pixel located 1 pixel left and above the pixel of interest 120a shown in FIG. The fourth pixel existing on one pixel of the numbered pixel is also investigated.

  Here, in this embodiment, the CPU 301 connects pixels existing in the vicinity of the pixel of interest to be investigated to the pixel of interest regardless of whether the investigation method of 4 neighborhood or 8 neighborhood is used. Is treated as a pixel. That is, in this embodiment, as illustrated in FIG. 5C, when the CPU 301 executes the labeling process on the pixels existing in the end column, the pixels existing in the end column and the start column are connected. It is configured to execute the labeling process.

  Next, labeling processing executed by the CPU 301 using the image processing program 304b in the present embodiment will be described with reference to FIGS. FIG. 6 is a flowchart showing the labeling process executed by the CPU 301.

  First, the CPU 301 determines whether or not the pixel value of the target pixel 120a is the second pixel value (step S1). In this process, the CPU 301 determines whether the pixel value of the target pixel 120a is the first pixel value or the second pixel value, and the target pixel 120a is a target of a label value adding process described later. It is determined whether or not.

  If it is determined in step S1 that the pixel value of the target pixel 120a is the second pixel value (YES), the CPU 301 executes label value addition processing (step S2). In this process, the CPU 301 executes a label value assigning process shown in FIG. 7, and assigns a label value to the target pixel 120a with reference to the label value given to the pixel connected to the target pixel 120a.

  After executing the process of step S2 or when determining that the pixel value of the target pixel 120a is the first pixel value in the process of step S1 (NO), the CPU 301 executes the process of step S3. To do. In the processing of step S3, the CPU 301 determines whether or not the target pixel 120a is present at the labeling processing end position 120b that is the right end and the lower end of the binarized image data 120 shown in FIG. 9A.

  If it is determined in step S3 that the target pixel 120a does not exist at the labeling process end position 120b (NO), the CPU 301 executes the target pixel movement process (step S4), and the process of step S1 is performed. Return. In this process, if the target pixel 120a exists in a position other than the end column, the CPU 301 moves the target pixel 120a by 1 pixel to the right in the horizontal direction. In this process, when the target pixel 120a exists in the end column, the CPU 301 moves the target pixel 120a by 1 pixel downward in the vertical direction and moves to the start column of the binarized image data 120.

  On the other hand, if it is determined in step S3 that the target pixel 120a is a pixel existing at the labeling process end position 120b (YES), the CPU 301 moves the target pixel to the labeling process start position (step S5). . In this process, since the CPU 301 has finished executing the label value adding process on the pixel having the second pixel value in the binarized image data 120, the left side of the binarized image data shown in FIG. The target pixel 120a is moved to the labeling processing start position 120c which is the upper end.

  Next, the CPU 301 executes a label value update process (step S6), and ends the labeling process. In this process, the CPU 301 executes the label value update process shown in FIG. 8 and assigns the same label value to each of the connected pixel values (hereinafter referred to as “labeled pixels”). To do.

  FIG. 7 is a flowchart showing the label value giving process executed in step S2 of the labeling process shown in FIG. In the present embodiment, the CPU 301 performs a survey of 8 neighborhoods of the pixel of interest 120a, but may be configured to conduct a survey of 4 neighborhoods.

  First, the CPU 301 determines in which column of the binarized image data 120 the pixel of interest 120a is present (step S11). In this process, the CPU 301 determines whether the target pixel 120a exists in the start column, the end column, or any other column of the binarized image data 120, and sets the position of the target pixel 120a. The processing to be executed is determined accordingly.

  If it is determined in step S11 that the target pixel 120a is present in the start column of the binarized image data (start column), the CPU 301 executes a start column neighborhood inspection process (step S12). In this processing, as shown in FIG. 5B, the CPU 301 determines that the pixels existing in the 0th and 1st positions connected to the target pixel 120a existing in the start column are labeled pixels. It is investigated whether it is.

  In the process of step S11, when it is determined that the target pixel 120a is present in the end column of the binarized image data (end column), the CPU 301 executes an end column neighborhood inspection process (step S13). In this processing, as shown in FIG. 5C, the CPU 301 determines that the pixels existing at the 0th, 1st, and 3rd positions connected to the target pixel 120a existing in the end column are respectively Check whether the pixel is a labeled pixel. Further, as shown in FIG. 5C, the CPU 301 also regards the pixels existing in the start row as the second and fourth positions as being connected to the target pixel 120a. It is investigated whether or not the pixel is attached.

  When it is determined in step S11 that the target pixel 120a is present in a column that is neither the start column nor the end column of the binarized image data (others), the CPU 301 performs the normal neighborhood inspection process. Execute (Step S14). In this processing, as shown in FIG. 5A, the CPU 301 investigates whether or not the pixels existing at positions 0 to 3 connected to the target pixel 120a are labeled pixels. .

  After executing any one of steps S12 to S14, the CPU 301 determines how many labeled pixels are connected to the target pixel 120a (step S15). In this process, the CPU 301 determines how many of the pixels connected to the target pixel 120a are labeled pixels, and gives the number of connected labeled pixels and each labeled pixel. The processing to be executed is determined according to the label value.

  If it is determined in step S15 that no labeled pixel is connected to the target pixel 120a (0), the CPU 301 assigns the label value currently held in the RAM 303 to the target pixel 120a. In step S16, a label value is assigned. Here, the label value is held in a predetermined area provided in the RAM 303 of the image processing apparatus 300, and the initial value is 1 at the start of the labeling process.

  Next, the CPU 301 increments the label value currently possessed (step S17), and ends the label value assigning process. In this process, the CPU 301 increments the label value possessed in the RAM 303, that is, adds one. With this process, the CPU 301 assigns a label value different from the label value assigned in the process of step S16 when the label value is assigned to the pixel not connected to the pixel to which the label value is assigned in the process of step S16. be able to.

  In the process of step S15, when it is determined that one labeled pixel is connected to the target pixel 120a and when it is determined that two or more labeled pixels to which the same label value is assigned are connected. (1), the CPU 301 executes the process of step S18. In the process of step S18, the CPU 301 assigns the same label value to the target pixel 120a as the label value assigned to the connected labeled pixels, thereby connecting the pixels having the second pixel value. Can be clear. After executing the process of step S18, the CPU 301 ends the label value giving process.

  In the processing of step S15, when it is determined that two or more labeled pixels each having a different label value are connected to the target pixel 120a (two or more), the CPU 301 pays attention to the minimum label value. This is applied to the pixel 120a (step S19). In this process, the CPU 301 first compares the label values assigned to each labeled pixel connected to the target pixel 120a. Then, the CPU 301 extracts from the result of comparing the smallest label value having the smallest value among the label values assigned to each labeled pixel, and assigns the extracted minimum label value to the target pixel 120a. A specific example of step S19 is shown in FIG. As shown in FIG. 9, the target pixel 120a is connected to a labeled pixel to which a label value “1” is assigned and a labeled pixel to which a label value “3” is assigned. In this case, the CPU 301 compares the label value of the labeled pixel to which the label value “1” is assigned with the labeled value to which the label value “3” is assigned, thereby labeling as the minimum label value. The value “1” is extracted, and the label value “1” is assigned to the target pixel 120a.

  After executing the process of step S19, the CPU 301 executes a lookup table update process (step S20) and ends the label value providing process. Here, the lookup table is stored in a predetermined area of the RAM 303, and as shown in FIG. 10A, an index number and a label value corresponding to the index number are stored. In the initial state, the lookup table matches the index number with the label value. In the present embodiment, the lookup table is configured to have up to index number “6”, but is not limited to this. The lookup table is configured to have an index number corresponding to the assigned label value when a label value larger than the label value “6” is given in the label value assigning process.

  In the look-up table update process, the CPU 301 first refers to the look-up table and specifies an index number having the same number as the label value determined to be larger than the minimum label value in the process of step S19. Next, the CPU 301 executes processing for updating the label value corresponding to the identified index number stored in the lookup table to the minimum label value.

  In the present embodiment, for example, when the target pixel 120a exists at the position shown in FIG. 9B, the CPU 301 first refers to the lookup table by the process of step S20, and reads the label value “ The index number “3” having the same number as “3” is specified. Next, the CPU 301 updates the label value “3” corresponding to the index number “3” stored in the lookup table to the label value “1” that is the minimum label value. Further, for example, when the target pixel 120a exists at the position illustrated in FIG. 9C, the CPU 301 performs label processing stored in the lookup table by the same processing as in the case of the label value “3” described above. The value “4” is updated to the label value “1”.

  In addition, when the target pixel exists at the position illustrated in FIG. 9D, the CPU 301 exists in the start column and the label value “5” of the labeled pixel that exists in the end column by the process of step S <b> 13. The label value “6” of the labeled pixel can be extracted. In the process of step S20, the CPU 301 updates the label value “6” to the label value “5” as in the case of the label value “3” described above. Thereby, in the label value update process described later, the CPU 301 connects the labeled pixel having the label value “5” present in the end column and the labeled pixel having the label value “6” present in the start column. As a result, the same label value can be given to both pixels.

  By executing the processing of step S19 and step S20 on the pixels connected to two or more labeled pixels among the pixels having the second pixel value, the CPU 301 can execute the processing shown in FIG. The updated look-up table shown in FIG.

  By executing the label value giving process shown in FIG. 7 for all the pixels having the second pixel value, the CPU 301 gives the label value to all the pixels having the second pixel value shown in FIG. Labeled image data 130 can be obtained. However, as shown in FIG. 9E, in the label-added image data 130, different label values are often assigned to the connected labeled pixels. In particular, in the label-added image data 130, the label value “6” is assigned to the pixel corresponding to the defect 110a, and the label value “5” is assigned to the pixel corresponding to the defect 110b. Therefore, in order to give the same label value to each of the connected labeled pixels, the CPU 301 executes a label value update process.

  FIG. 8 is a flowchart showing the label value update process executed in step S6 of the labeling process shown in FIG.

  First, the CPU 301 determines whether or not the target pixel 120a is a labeled pixel (step S31). In this process, if it is determined that the target pixel 120a is not a labeled pixel (NO), the CPU 301 determines whether or not the target pixel 120a is connected to the labeled pixel (step S32). In these processes, the CPU 301 determines whether or not to execute a process for updating a later-described label value for the target pixel 120a and the pixels connected to the target pixel 120a.

  When it is determined that the target pixel 120a is a labeled pixel (YES in step S31), or when it is determined that the target pixel 120a is connected to the labeled pixel (YES in step S32), the CPU 301 performs step The process of S33 is executed. In the processing of step S33, the CPU 301 extracts the target pixel 120a and the label value given to the pixel connected to the target pixel 120a, and acquires the label value from the updated lookup table shown in FIG. Update the label value.

  The details of step S33 will be described. First, the CPU 301 extracts the label value assigned to the labeled pixel, and specifies the index number having the same number as the extracted label value. Next, the CPU 301 refers to the updated look-up table shown in FIG. 10B, acquires a label value corresponding to the specified index number, and assigns the acquired label value to the labeled pixel. The label value of the attached pixel is updated.

  For example, when the target pixel 120a has the label value “3”, the CPU 301 first extracts the label value “3” of the labeled pixel and identifies the index number “3”. Next, the CPU 301 refers to the updated look-up table, acquires the label value “1” corresponding to the identified index number “3”, and assigns the label value “1” to the labeled pixel, thereby The label value of the attached pixel is updated. When the pixel connected to the target pixel 120a has the label value “1”, the CPU 301 first extracts the label value “1” of the connected labeled pixel and specifies the index number “1”. To do. Next, the CPU 301 refers to the updated lookup table, acquires the label value “1” corresponding to the identified index number “1”, and assigns the label value “1” to the connected labeled pixels. As a result, the label value is updated. With this process, the CPU 301 updates the label value “6” to the label value “5”. That is, the CPU 301 can assign the same label value “5” to both the pixels corresponding to the defect 110a existing in the start column and the pixel corresponding to the defect 110b existing in the end column. .

  After executing the process of step S33 or when determining in step S32 that the target pixel 120a is not connected to the labeled pixel (NO), the CPU 301 executes the process of step S34. In the process of step S34, the CPU 301 determines whether or not the target pixel 120a is present at the labeling process end position 120b.

  If it is determined in step S34 that the target pixel 120a does not exist at the labeling process end position 120b (NO), the CPU 301 executes the target pixel movement process (step S35), and the process of step S31 is performed. Return. In this process, the CPU 301 executes a process similar to the process in step S4 of the labeling process (see FIG. 6).

  On the other hand, if it is determined in step S34 that the target pixel 120a is a pixel existing at the labeling processing end position 120b (YES), the CPU 301 has finished updating the label value of the binarized image data 120. The label value update process is terminated.

  By executing the label value update process shown in FIG. 8, the CPU 301 updates the label value-updated image data 140 in which the same label value is assigned to each of the connected labeled pixels shown in FIG. Can be obtained.

  As described above, the image processing apparatus 300 according to the present embodiment executes the end-sequence neighborhood inspection process (see FIG. 7) when the target pixel 120a exists in the end sequence, and uses the survey result. Each process (see steps S16, S18, and S19 in FIG. 7) for assigning a label value is executed. In the label value assigning process, the label value updating process (see FIG. 8) is executed with reference to the updated post-update lookup table (see FIG. 10B). Thereby, even if the image processing apparatus 300 divides one defect 11a as the defect 110a and the defect 110b at both ends of the polar coordinate image data 110 according to the start coordinates of the polar coordinate conversion process, the defect 110a and the defect 110b are divided. Can be determined to be connected. Then, the image processing apparatus 300 assumes that the defect 110a existing at one end 120d of the binarized image data 120 and the defect 110b existing at the other end 120e are one defect 11a, and the defect 110a and the defect 110b. Can be given the same label value. That is, the image processing apparatus 300 assigns the same label value to the divided defect 110a and defect 110b without performing complicated image processing, so that the image processing can be simplified and increased in efficiency. It becomes possible.

  Then, the inspection apparatus 1 extracts and analyzes the feature amount of the defect 11a from the updated label value image data 140 on which the labeling process has been performed. For this reason, the inspection apparatus 1 can execute the defect determination assuming that the defect 110a and the defect 110b divided by the polar coordinate conversion process are the same defect 11a. As a result, the inspection apparatus 1 can accurately perform the defect determination in the defect determination of the inspection object 11 without causing the problem of the feature amount analysis process caused by the polar coordinate conversion process.

  In the present embodiment, the image processing apparatus 300 is used as the inspection apparatus 1 to extract the defect of the inspection object 11 as an example. However, the present embodiment is applied to a so-called machine vision or is not limited thereto. It can be applied to other fields. For example, the image processing apparatus 300 can be used as a process for inspecting the quality of a produced article (product) in a production apparatus for optical elements such as lenses. That is, the above-described image processing method is executed and an article is inspected by an inspection apparatus including a manufacturing process for manufacturing an article, an imaging apparatus that images the article, and an image processing apparatus that processes an image captured by the imaging apparatus. You may use for the production method which performs an inspection process. By using the image processing method described above, an article can be effectively produced in the article production method. In short, the present technology can be applied to any image processing as long as the image processing extracts a defect from image data centered on a point-symmetric point.

DESCRIPTION OF SYMBOLS 1 ... Inspection apparatus: 11 ... Inspected object: 11a ... Defect: 11b ... Point symmetry point: 11c ... Start position: 11d ... End position: 12 ... Positioning member (holder): 13 ... Light source: 14 ... Imaging apparatus (camera) ): 100: Image data: 110: Image data subjected to polar coordinate conversion processing (polar coordinate image data): 120 ... Image data subjected to binarization processing (binarized image data): 120d: One end: 120e: Other end : 300 ... Image processing apparatus: 301 ... Control unit (CPU): 303 ... Storage unit (RAM): 304 ... Recording medium (HDD): 304b ... Image processing program

Claims (7)

A storage unit capable of storing image data acquired by imaging the inspection object;
Performing a polar coordinate conversion process from a circumferential start position starting from a rotation center point of the measurement target region in the image data stored in the storage unit to a circumferential end position connected to the start position, Performing binarization processing to binarize a pixel constituting image data subjected to polar coordinate conversion processing into a first pixel value and a second pixel value with a predetermined threshold as a boundary; When pixels converted into either one of the first pixel value or the second pixel value are connected by the conversion process, the same label value is assigned to each of the connected pixels. A control unit that executes a labeling process to be provided,
In the labeling process, the controller assigns the label value to each of one end corresponding to the start position and the other end corresponding to the end position of the image data that has been subjected to the binarization process. If present, the same label value is given as if the pixels existing at the one end and the other end are connected,
An image processing apparatus. A positioning member for positioning the inspection object;
A light source for irradiating the object to be inspected;
An imaging device that images the inspection object and acquires the image data;
An image processing apparatus according to claim 1,
Using the image data to which the label value is given by the image processing device, it is determined whether or not the inspection object has a defect,
Inspection apparatus characterized by that. Polar coordinates from the start position in the circumferential direction starting from the center point of rotation of the measurement target region in the image data of the inspection object stored in the storage unit to the end position in the circumferential direction connected to the start position A polar coordinate conversion process for executing the conversion process;
The control unit binarizes the pixels constituting the image data that has been subjected to the polar coordinate conversion process in the polar coordinate conversion process step into a first pixel value and a second pixel value with a predetermined threshold as a boundary. A binarization process for executing the binarization process;
In the image data in which the control unit has performed the binarization process in the binarization process step, a pixel converted into one of the first pixel value and the second pixel value is A labeling process that executes a labeling process that gives the same label value to each of the connected pixels when connected, and
Pixels to which the control unit assigns the label value to each of one end corresponding to the start position and the other end corresponding to the end position of the image data subjected to the binarization process in the labeling process step If there is, the end processing to give the same label value as the connection between the one end and both pixels existing at the other end,
An image processing method. In the labeling process, when the control unit connects at least one labeled pixel to which the label value is given to the pixel to which the label value is given, the connected labeling A label value assigning process for assigning the smallest minimum label value among the label values given to the pixels to the pixel to which the label value is given;
In the labeling process, when the control unit is connected to the pixel to which the label value is applied and at least one of the labeled pixels is connected, the control unit is applied to the connected labeled pixels. A label value update process for updating the label value different from the minimum label value to the minimum label value;
In the edge processing, the control unit assigns the label value when there is a pixel to which the label value is applied at each of the one end and the other end of the image data subjected to the binarization processing. Process, and the label value update process, to give the same minimum label value to both pixels existing at the one end and the other end,
The image processing method according to claim 3.   An image processing program for causing a computer to execute each step according to claim 3 or 4.   A computer-readable recording medium on which the image processing program according to claim 5 is recorded. A manufacturing process for manufacturing the article;
An inspection step of inspecting the article by executing the image processing method according to claim 4 by an inspection apparatus including an imaging device that images the article and an image processing device that processes an image captured by the imaging device. And with,
A production method characterized by that.

Images