JP6244260B2 - Defect analysis apparatus and defect analysis method - Google Patents

Description

  The present invention relates to a defect analysis apparatus and a defect analysis method for analyzing defects generated in a long product in a manufacturing process.

  In the manufacturing process of long products such as metal plates such as rolled steel plates, paper, films, metal thin plates, metal foils, etc., the surface of the product is imaged with a camera and the resulting images are processed to obtain surface defects. Inspection (scratches, dirt, foreign matter, etc.) is performed.

  In a long product manufacturing process, when a product is conveyed by a roller, defects that are periodically distributed in the product due to foreign matter adhering to the roller (hereinafter referred to as “periodic defects”) may occur. is there. Identifying such periodic defects is very useful for inspecting rollers.

  Patent Document 1 describes a periodic defect inspection apparatus that extracts periodic defects from detected defects. In this periodic defect inspection apparatus, a group of defects detected at approximately equal intervals in the longitudinal direction at positions where defects are present in the steel sheet is determined to be a periodic defect group.

Japanese Unexamined Patent Publication No. 2011-242318

  However, in the periodic defect inspection apparatus described in Patent Document 1, if a periodic defect intersects a periodic defect or a large number of discrete defects occur in a wide range, the periodic defect is accurately identified. Difficult to do.

  This invention is made | formed in view of such a situation, The main objective is to provide the defect analysis apparatus and defect analysis method which can solve the said subject.

In order to solve the above-described problem, a defect analysis apparatus according to one aspect of the present invention is a defect analysis apparatus that analyzes defects distributed in a target product, and is based on a position of each defect in the target product in a predetermined direction. The defect group distinguishing means for distinguishing the first defect group periodically distributed from the second defect group different from the first defect group and the second defect group are distinguished by the defect group distinguishing means Period detecting means for detecting a distribution period of defects included in the first defect group, and the defect group distinguishing means clusters each defect into a plurality of defect groups based on the position of each defect in the target product. Clustering means, feature information extracting means for extracting feature information of the defect group, and feature information extracting means extracted for each defect group obtained by clustering in the clustering means. Was based on the feature information comprises a classification means for classifying the defect groups and the second defect group and the first defect group.

  In the above aspect, the feature information may include a length of a defect group in each of the predetermined direction and a direction orthogonal to the predetermined direction.

  Further, in the above aspect, the clustering means may include a first clustering means for clustering defects based on a first position condition, and a second clustering for defects based on a second position condition different from the first position condition. Clustering means.

  Further, in the above aspect, the first position condition is that a distance between a plurality of defects is within a predetermined distance, and the second position condition is that the plurality of defects exist in a region long in the predetermined direction. It may be.

  In the above aspect, the second clustering unit may be configured to cluster the remaining defects obtained by excluding the defects clustered by the first clustering unit from the defects existing in the target product.

  In the above aspect, the period detection unit may be configured to detect the shortest distance between adjacent defects in the first defect group as the distribution period.

  Further, in the above aspect, the defect analysis apparatus may be configured such that, in the first defect group distinguished from the second defect group by the defect group distinguishing unit, a plurality of defect distances are within a predetermined set distance. The apparatus further comprises third clustering means for clustering the plurality of defects as a third defect group, wherein the period detection means uses the third defect group obtained by the third clustering means as one defect belonging to the first defect group. As described above, the distribution period may be detected.

  In the above aspect, the period detection unit may be configured to detect the distribution period using a representative value of the position of the third defect group as the position of the one defect.

The defect analysis method according to one aspect of the present invention is a defect analysis method for analyzing defects distributed in a target product, wherein the defect analysis method periodically distributes in a predetermined direction based on the position of each defect in the target product. Distinguishing between one defect group and a second defect group different from the first defect group, and detecting a distribution period of defects included in the first defect group distinguished from the second defect group If, have a, the distinguishing step is based on the position of each defect in the products, the steps of clustering the defects to the plurality of defect groups, for each defect group obtained by clustering of defects group wherein Extracting information, and classifying the defect group into the first defect group and the second defect group based on the extracted feature information .

  According to the defect analysis apparatus and the defect analysis method according to the present invention, it is possible to accurately detect the period of a periodic defect.

The schematic diagram which shows the whole structure of the defect analysis system containing the defect analyzer which concerns on embodiment. The block diagram which shows the structure of the defect analyzer which concerns on embodiment. The flowchart which shows the procedure of the defect analysis process by the defect analyzer which concerns on embodiment. The flowchart which shows the procedure of a clustering process. The schematic diagram for demonstrating a 1st position condition and a 2nd position condition. The flowchart which shows the procedure of a period detection process. The schematic diagram explaining the outline | summary of a reclustering process. The schematic diagram explaining the outline | summary of the process of step S206 of FIG.

  Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings.

[Configuration of defect analyzer]
FIG. 1 is a schematic diagram showing an overall configuration of a defect analysis system including a defect analysis apparatus according to the present embodiment. The defect analysis system 100 includes a CCD camera 200, a defect detection device 300, and the defect analysis device 1. The defect analysis system 100 is for analyzing a defect of a long target product S such as a rolled steel sheet in a manufacturing process. In the figure, the target product S is conveyed in the Y direction.

  The CCD camera 200 is disposed above the conveyance path of the target product S and images the surface (upper surface) of the target product S.

  The defect detection apparatus 300 is for detecting defects on the surface of the target product S by image processing. The defect detection device 300 is connected to the CCD camera 200 and receives image data output from the CCD camera 200. The defect detection apparatus 300 detects a defect image included in the obtained image, and detects position information of the defect image (X direction, that is, the width direction of the target product S, and Y direction, that is, the length of the target product S). Direction coordinates), and a defect image which is a partial image is cut out. The defect position information and the defect image are output from the defect detection apparatus 300.

  The defect analysis apparatus 1 is connected to the defect detection apparatus 300. The defect analysis apparatus 1 takes in positional information and defect images of defects output from the defect detection apparatus 300 and analyzes defects based on these data.

  FIG. 2 is a block diagram showing the configuration of the defect analysis apparatus 1. The defect analysis apparatus 1 is realized by a computer 10. As shown in FIG. 2, the computer 10 includes a main body 11, an input unit 12, and a display unit 13. The main body 11 includes a CPU 111, a ROM 112, a RAM 113, a hard disk 115, a reading device 114, an input / output interface 116, an image output interface 117, and a communication interface 118. The CPU 111, the ROM 112, the RAM 113, the hard disk 115, the reading device 114, an input device 114, The output interface 116, the image output interface 117, and the communication interface 118 are connected by a bus.

  The CPU 111 can execute a computer program loaded in the RAM 113. Then, when the CPU 111 executes a defect analysis program 110 that is a computer program for defect analysis, the computer 10 functions as the defect analysis apparatus 1.

  The ROM 112 is configured by a mask ROM, PROM, EPROM, EEPROM, or the like, and stores a computer program executed by the CPU 111, data used for the same, and the like.

  The RAM 113 is configured by SRAM, DRAM, or the like. The RAM 113 is used to read the defect analysis program 110 recorded on the hard disk 115. Further, when the CPU 111 executes a computer program, it is used as a work area for the CPU 111.

  The hard disk 115 is installed with various computer programs to be executed by the CPU 111 such as an operating system and application programs, and data used for executing the computer programs. The defect analysis program 110 is also installed on the hard disk 115.

  The hard disk 115 is installed with an operating system such as Windows (registered trademark) manufactured and sold by Microsoft Corporation. In the following description, it is assumed that the defect analysis program 110 according to the present embodiment operates on the operating system.

  The input / output interface 116 is, for example, a serial interface such as USB, IEEE1394, or RS-232C, a parallel interface such as SCSI, IDE, or IEEE1284, and an analog interface including a D / A converter, an A / D converter, and the like. It is configured. An input unit 12 including a keyboard and a mouse is connected to the input / output interface 116, and the user can input data to the computer 10 by using the input unit 12.

  The image output interface 117 is connected to the display unit 13 constituted by an LCD, a CRT, or the like, and outputs a video signal corresponding to the image data given from the CPU 111 to the display unit 13. The display unit 13 displays an image (screen) according to the input video signal.

  The communication interface 118 transmits and receives data according to a predetermined communication protocol. The above-described defect detection device 300 is connected to the communication interface 118.

[Operation of defect analyzer]
Hereinafter, the operation of the defect analysis apparatus 1 according to the present embodiment will be described.

  The defect analysis apparatus 1 executes processing as described below, identifies a periodic defect from the defects of the target product S, and detects the period.

  FIG. 3 is a flowchart showing a procedure of defect analysis processing by the defect analysis apparatus 1 according to the present embodiment.

  The defect analysis process includes a defect group discrimination process S1 and a cycle detection process S2. The defect group discrimination process S1 and the cycle detection process S2 are executed by the CPU 111.

  The defect group distinguishing process S1 includes, for each defect of the target product S, a periodic defect group (first defect group) that is periodically distributed in the Y direction (length direction) and a defect group other than the periodic defect group (first defect group). 2 defect group). The defect group discrimination process S1 includes defect coordinate reading (step S11), clustering process (step S12), cluster feature information extraction (step S13), and cluster classification (step S14).

  The period detection process S2 is a process for detecting a period in which each defect is distributed with respect to the first defect group distinguished from the second defect group by the defect group distinction process S1, that is, the periodic defect group.

  Hereinafter, the defect analysis process will be described in detail.

  First, the CPU 111 reads out defect coordinates fetched from the defect detection apparatus 300, that is, defect position information (step S11). In the following description, the defect coordinate of the xth defect is denoted as P (x). The number of the cluster to which the xth defect belongs is denoted as C (x).

  When reading of all defect coordinates is completed, the CPU 111 executes clustering processing (step S12).

  FIG. 4 is a flowchart showing the procedure of the clustering process.

  In the clustering process S12, the CPU 111 first initializes variables n and k to 0 (step S101), and initializes variable i to 1 (step S102). Note that n is a variable indicating the number of a defect as a reference (hereinafter referred to as “reference defect”), and i is a variable indicating the number of a defect to be compared (hereinafter referred to as “comparative defect”).

  The CPU 111 selects a reference defect coordinate P (n) and a comparison defect coordinate P (i) (step S103).

  Next, the CPU 111 determines whether or not the cluster number C (i) of the comparative defect is 0, which is an initial value (step S104). If C (i) is other than 0, it means that the comparison defect has already been clustered. In this case (NO in step S104), the CPU 11 moves the process to step S111, and determines whether i is the final (last defect number). If i is not final (NO in step S111), CPU 111 increments i (step S112) and returns the process to step S103.

  If C (i) is 0 in step S104 (YES in step S104), CPU 111 determines whether or not the first position condition is satisfied (step S105). FIG. 5 is a schematic diagram for explaining the first position condition and the second position condition. The first position condition is a condition that the distance K between adjacent defects is within a predetermined distance. That is, the CPU 111 calculates the distance between P (n) and P (i), and determines whether the distance is equal to or less than a predetermined distance. When the first position condition is satisfied, a comparative defect exists in a circle having a radius of a predetermined distance centering on the reference defect. That is, the reference defect and the comparative defect are close to each other. In the clustering process S12, a plurality of defects close to each other (hereinafter referred to as “dense defect group”, and defects constituting the dense defect group are referred to as “dense defect”) are clustered into the same cluster.

  The predetermined distance of the first position condition is preferably determined as a value smaller than the minimum value of the periodic defect assumed, and more preferably half or less of the minimum value of the period. The lower limit value of the predetermined distance is a minimum value of the length of the defect image in the X direction or the Y direction. This makes it possible to cluster only dense defects without clustering periodic defects.

  When the first position condition is satisfied in step S105 (YES in step S105), the CPU 111 determines whether or not the cluster number C (n) of the reference defect is 0 (step S106). When the cluster number C (n) of the reference defect is other than 0 (NO in step S106), the reference defect has already been assigned to any cluster. In this case, the CPU 111 sets the cluster number C (i) of the comparative defect to the same value as the cluster number C (n) of the reference defect (step S107). Thereby, the comparative defect is clustered into the same cluster as the reference defect.

  On the other hand, when the cluster number C (n) of the reference defect is 0 (YES in step S106), the reference defect is not yet clustered in any cluster. Therefore, the CPU 111 increments k (step S108), and sets the cluster numbers C (n) and C (i) of both the reference defect and the comparative defect to the value of k (step S109). Thereby, the comparative defect and the reference defect are clustered into a new cluster.

  After step S107 or S109, the CPU 111 moves the process to step S111.

  If the first position condition is not satisfied in step S105 (NO in step S105), the CPU 111 determines whether or not the second position condition is satisfied for P (n) and P (i). (Step S110).

  The second position condition will be described with reference to FIG. The second position condition is a condition that there are a plurality of defects in a rectangular area that is long in the Y direction. In this rectangular area, the length W in the X direction is a value including at least the meandering range in the conveyance of the target product S, and the length in the Y direction is a value including at least the length of the outermost coil of the target product S. Is done. The CPU 111 determines whether P (n) and P (i) are included in the rectangular area. When the second position condition is satisfied, the reference defect and the comparative defect are aligned along the conveyance direction of the target product S. In the clustering process S12, a plurality of defects arranged in the transport direction of the target product S are clustered into the same cluster.

  If the second position condition is satisfied in step S110 (YES in step S110), CPU 111 executes the processing after step S106.

As described above, a defect that has been clustered once is excluded from the objects of subsequent clustering. In other words, once a defect has been assigned a cluster number, the cluster number is not changed thereafter. Further, the clustering based on the first position condition is performed prior to the clustering based on the second position condition. For this reason, even when both the first position condition and the second position condition are satisfied, clustering is performed on the defect based on the first position condition, and clustering based on the second position condition is not performed. Therefore, even when other defect groups such as dense defect groups are distributed so as to overlap with the periodic defect group, clustering can be performed by distinguishing the periodic defect group from the other defect group.

  By repeating the above processing until i becomes final, clustering can be performed for the reference defect P (n) and all the defects P (i).

  In step S111, when i is final (YES in step S111), the CPU 111 determines whether n is final (last defect number) (step S113). If n is not final (NO in step S113), CPU 111 increments n (step S114) and returns the process to step S102.

  By repeating the above processing until n becomes final, all defects can be selected as the reference defect P (n) and clustering can be performed. Although omitted in the above description, clustering is not performed when n and i have the same value.

  When n is final in step S113 (YES in step S113), the CPU 111 ends the clustering process and returns the process to the main routine.

  The continuation of the defect analysis process will be described with reference to FIG. 3 again. When the clustering process S12 ends, the CPU 111 extracts feature information from each cluster (step S13). The feature information is information indicating features related to the form of the cluster, and includes the length in the X direction, the length in the Y direction, and the density (the number of defects present per unit area). In addition, when the defect detection apparatus 300 outputs information on a defect rank (importance) and a defect type (insect, depression by pressing, etc.), it is also possible to include such information in the feature information. Further, the area of the cluster, the aspect ratio, and the number of defects included in the cluster can be included in the feature information. Note that the feature information may not include all of the information described above, but includes at least the length in the X direction and the length in the Y direction.

  Next, the CPU 111 performs cluster classification (step S14). This cluster classification is performed using the feature information of each cluster extracted in step S13. In this processing, each cluster is divided into “long streaks”, “transfer continues (hereinafter referred to as“ transfer defects ”)”, “long in the X direction”, “aggregate”, “ Classify as "Expanding".

  Among the above-mentioned “long streaks”, “transfer defects”, “long in the X direction”, “lumps”, “those that spread all over”, “long streaks” and “transfer defects” are periodic defect groups. (First defect group). On the other hand, “long in the X direction”, “globular”, and “wide spread” are defect groups (second defect group) different from the periodic defect group. As described above, the first defect group and the second defect group are distinguished by the classification of the clusters.

  After the defect group distinction process as described above, the CPU 111 executes a cycle detection process (step S2). This period detection process is a process of calculating a distribution period (pitch) in the defect transport direction from each cluster of the first defect group.

  FIG. 6 is a flowchart illustrating the procedure of the cycle detection process.

  In the cycle detection process S2, first, the CPU 111 executes a reclustering process (step S201). In the reclustering process, clustering is performed again for each defect belonging to the cluster of the periodic defect group.

  FIG. 7 is a schematic diagram for explaining the outline of the reclustering process. As shown in FIG. 7, the periodic defect may include a plurality of locally dense defects. For example, such a phenomenon is observed when a defect that is originally a defect is recognized as a plurality of defects as a result of the occurrence of a portion with different shades in the image of the defect. Such a phenomenon is also observed when another defect is present near one defect. In the reclustering process, such dense defects are recognized as one cluster.

  In the reclustering process, it is determined whether or not the distance between defects is equal to or less than a predetermined set distance for all combinations of two defect coordinates. This set distance is set to a value smaller than the assumed minimum value of the periodic defect period (for example, the circumferential length of the smallest conveying roller included in the manufacturing process). The lower limit value of the set distance is the minimum value of the length of the defect image in the X direction or Y direction. This makes it possible to cluster only dense defects without clustering periodic defects.

  Next, the CPU 111 sets the representative point coordinates of the reclustered defect group (hereinafter referred to as “third defect group”) (step S202). The setting of the representative point coordinates will be described with reference to FIG. In the present embodiment, the center of gravity position of the cluster is set as the representative point coordinates. In the following processing, the reclustered cluster is treated as one defect, and the representative point coordinates of the cluster are treated as the coordinates of the defect. It is also possible to set one point in the cluster other than the center of gravity position as the representative point coordinates such as the cluster center position, the first defect coordinate in the Y direction of the cluster, and the last defect coordinate in the Y direction.

  Next, the CPU 111 initializes the variable j to 1 and the variable B to 0 (step S203). Note that j is a variable indicating a defect number, and B is a variable indicating the number of defects detected that are not periodic defects.

  The CPU 111 selects the first two defects, that is, P (0) and P (1), and calculates a distance D0 between them (step S204). Further, the CPU 111 selects P (j) and P (j + 1), and calculates a distance D (j) between them (step S205).

  Next, the CPU 111 determines whether D (j) / D0 or D0 / D (j) is an integer (step S206). FIG. 8 is a schematic diagram for explaining an overview of the processing in step S206. In periodic defects, the distance between adjacent defects is constant. That is, the values of D (j) / D0 and D0 / D (j) should be 1. However, due to the influence of the imaging state of the CCD camera 200, the surface state of the target product S, the state of the surrounding environment of the target product S, etc., some defect images are blurred or blurred and are not recognized as defects. (Defects indicated by broken lines in the figure). When defect recognition failure occurs in this way, D (j) becomes an integer multiple of twice or more of D0, or D0 becomes an integer multiple of twice or more of D (j). In consideration of such a case, in step S206, not only D (j) / D0 or D0 / D (j) is 1 but also an integer is recognized as a periodic defect.

  More specifically, in the process of step S206, it is determined whether D (j) / D0 or D0 / D (j) is an integer ± predetermined value (for example, 0.1). This takes into account the error component of defect position detection caused by imaging or image processing, and allows a certain range centered on an integer to prevent detection of periodic defects due to overly strict judgment It is to do.

  As described above, in this embodiment, the distribution cycle of periodic defects is obtained regardless of the roll size in the manufacturing process. By doing in this way, even when the distribution cycle of periodic defects changes due to processing or the like in the manufacturing process of the target product S, the cycle can be accurately detected.

  In step S206, when D (j) / D0 or D0 / D (j) is an integer, the CPU 111 determines that P (j + 1) is a periodic defect, and when it is not an integer, P (j + 1) ) Is not a periodic defect.

  If D (j) / D0 or D0 / D (j) is an integer in step S206 (YES in step S206), CPU 111 moves the process to step S209.

  If D (j) / D0 or D0 / D (j) is not an integer in step S206 (NO in step S206), CPU 111 increments B (step S207) and determines whether B is greater than 2 or not. (Step S208). This process is to check whether or not the number of defects that are not periodic defects exceeds the allowable number (2 in this embodiment). That is, if the detection number B of defects that are not periodic defects is 2 or less, the cycle detection process is continued as it is, and if the detection number B of defects that are not periodic defects exceeds 2, the cycle detection process ends. Is done.

  In step S208, when B is 2 or less (NO in step S208), the CPU 111 moves the process to step S211 and sets the minimum value of the distance D calculated so far as a cycle (step S211). The cycle detection process ends.

  In step S208, if B is greater than 2 (YES in step S208), CPU 111 moves the process to step S209.

  In step S209, the CPU 111 determines whether j is final (the number of the last previous defect). If j is not final (NO in step S209), CPU 111 increments j (step S210) and returns the process to step S205.

  In step S209, if j is final (YES in step S209), the CPU 111 moves the process to step S211, sets the minimum value of the distance D calculated so far as a period (step S211), and performs a period detection process. Exit. When the cycle detection process ends, the defect analysis process ends.

  As described above, the first defect group (periodic defect group) is detected separately from the second defect group (defect group different from the periodic defect group), and the period is detected for the first defect group. Therefore, even if the second defect group exists so as to overlap the first defect group, it is possible to accurately detect the period by eliminating the influence of the second defect group.

(Evaluation test)
The defect analyzer described in the above embodiment was actually created and its performance was evaluated. The following table shows the result of detecting the distribution period of periodic defects by inputting defect coordinates obtained in a certain period by the defect detection apparatus to the defect analysis apparatus. One row of the table corresponds to one cluster. The label in the table is a label attached to the cluster, and is information for identifying each cluster. The Y coordinate value in the table is the Y direction component of the cluster representative point coordinate, and the X coordinate value is the X direction component of the cluster representative point coordinate. The number indicates the number of defects included in the cluster, the pitch indicates the detected distribution period, and the roll diameter indicates the roll diameter causing the periodic defect calculated from the pitch.

  The production equipment subjected to the evaluation test is provided with rolls having typical dimensions and diameters of 250 mm, 100 mm, and 50 mm, respectively. From the test results, it can be seen that the roll causing the defect can be accurately identified. Note that the roll diameters calculated to be about 592 mm and about 296 mm are the results of changes in the period of defects caused by rolls having actual diameters of 100 mm and 50 mm, respectively, as the target product S was extended in the Y direction. Generally agrees with Thus, it can be seen that this method can detect the period with sufficient accuracy for any periodic defect.

(Other embodiments)
In the above-described embodiment, the configuration of clustering the dense defect group (second defect group) in the clustering process has been described. However, the present invention is not limited to this. The second defect group may include a defect group other than the dense defect group. In the clustering process, it is possible to cluster a defect group which is different from the dense defect group and which is different from the periodic defect group. For example, it is possible to cluster the defect group existing in the region by using the position condition that the region exists in the region elongated in the X direction.

  Further, in the above-described embodiment, the configuration has been described in which defects are clustered, and after each cluster is classified, the period of periodic defects is detected. However, the present invention is not limited to this. Most of the defects clustered by the second position condition are periodic defects, and most of the defects clustered by the first position condition are defects different from the periodic defect. That is, it can be seen that the first defect group and the second defect group are distinguished by clustering. From this, it is also possible to detect the period of the clusters clustered using the second position condition without classifying the clusters.

  In the above-described embodiment, the configuration in which the period of the periodic defect is detected regardless of the size of the roll in the manufacturing process has been described. However, the present invention is not limited to this. For example, by using the smallest roll circumference in the manufacturing process, it is judged whether the distance between adjacent defects is an integral multiple of the circumference, thereby judging whether it is a periodic defect. It is also possible to detect the period of the defect based on the size of the roll in the manufacturing process by setting the shortest distance among the distances between the defects determined as defects as the period.

  In the above-described embodiment, the configuration in which the defects included in the cluster of the periodic defect group are reclustered by the reclustering process has been described. However, the present invention is not limited to this. It is also possible to adopt a configuration that does not perform re-clustering as described above. Further, after cluster classification, it is also possible to cancel a cluster determined as a periodic defect group and perform clustering again. At this time, clustering may be performed so that the periodic defect and other defects are further accurately separated. In order to simplify the management of the cluster, the center of gravity of the cluster, the center position of the cluster, the position of the tip or end defect in the Y direction in the cluster, etc. can be set as the representative point coordinates of the cluster. .

  In the above-described embodiment, the defect analysis process is executed by executing the defect analysis program by the CPU. However, the present invention is not limited to this. It is also possible to adopt a configuration in which the period of a periodic defect is detected by an application specific integrated circuit (ASIC) or a field-programmable gate array (FPGA) that can execute the same defect analysis processing as described above.

  In the above-described embodiment, the configuration in which an image is obtained by the CCD camera 200 that is an area sensor has been described. However, the present invention is not limited to this. The line sensor is arranged so as to extend in the width direction of the target product S, and while the target product S and the line sensor are relatively moved in the longitudinal direction of the target product S, the pixel sensor is continuously scanned by the line sensor. It is also possible to generate an image by combining these pixel column data.

  The defect analysis apparatus and the defect analysis method of the present invention are useful as a defect analysis apparatus and a defect analysis method for analyzing defects generated in a long product in a manufacturing process.

DESCRIPTION OF SYMBOLS 1 Defect analysis apparatus 10 Computer 12 Input part 13 Display part 100 Defect analysis system 200 Camera 300 Defect detection apparatus 110 Defect analysis program 111 CPU
115 Hard Disk 116 Input / Output Interface 117 Image Output Interface 118 Communication Interface

Claims (9)

A defect analyzer for analyzing defects distributed in a target product,
A defect group discriminating means for discriminating between a first defect group periodically distributed in a predetermined direction and a second defect group different from the first defect group based on the position of each defect in the target product;
A period detecting means for detecting a distribution period of defects included in the first defect group distinguished from the second defect group by the defect group distinguishing means;
Equipped with a,
The defect group distinguishing means includes
Clustering means for clustering each defect into a plurality of defect groups based on the position of each defect in the target product;
For each defect group obtained by clustering in the clustering means, feature information extraction means for extracting feature information of the defect group;
Classification means for classifying the defect group into the first defect group and the second defect group based on the feature information extracted by the feature information extraction means;
Comprising
Defect analyzer. The feature information includes lengths of defect groups in each of the predetermined direction and a direction orthogonal to the predetermined direction.
The defect analysis apparatus according to claim 1 . The clustering means includes
First clustering means for clustering defects according to a first position condition;
Second clustering means for clustering defects according to a second position condition different from the first position condition;
Having
The defect analysis apparatus according to claim 1 or 2 . The first position condition is that a distance between a plurality of defects is within a predetermined distance,
The second position condition is that a plurality of defects exist in a region long in the predetermined direction.
The defect analysis apparatus according to claim 3 . The second clustering unit is configured to cluster the remaining defects obtained by excluding the defects clustered by the first clustering unit from the defects existing in the target product.
The defect analysis apparatus according to claim 3 or 4 . The period detection unit is configured to detect the shortest distance between adjacent defects in the first defect group as the distribution period.
Defect analyzer according to any one of claims 1 to 5. In the first defect group distinguished from the second defect group by the defect group distinguishing means, the plurality of defects are clustered as a third defect group when the distance between the plurality of defects is within a predetermined set distance. A third clustering means;
The period detection unit is configured to detect the distribution period by using the third defect group obtained by the third clustering unit as one defect belonging to the first defect group.
Defect analyzer according to any one of claims 1 to 6. The period detection means is configured to detect the distribution period with the representative value of the position of the third defect group as the position of the one defect.
The defect analysis apparatus according to claim 7 . A defect analysis method for analyzing defects distributed in a target product,
Distinguishing between a first defect group periodically distributed in a predetermined direction based on a position of each defect in the target product and a second defect group different from the first defect group;
Detecting a distribution period of defects included in the first defect group distinguished from the second defect group;
I have a,
The step of distinguishing includes
Clustering each defect into a plurality of defect groups based on the position of each defect in the target product;
For each defect group obtained by clustering, extracting feature information of the defect group;
Classifying the defect group into the first defect group and the second defect group based on the extracted feature information;
Having
Defect analysis method.