JP2006189293A - Inspection method and device of striped irregular defect - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an inspection method and a device of a striped irregular defect capable of acquiring high inspection performance without being influenced by the width in a short side direction of the striped irregular defect. <P>SOLUTION: This inspection method of the striped irregular defect has a noise removal process relative to an input image; a smoothing process for applying a smoothing filter in the direction of the striped irregular defect to be detected relative to a noise removed image; a primary differential process for applying a primary differential filter in the three directions excluding the defect direction relative to a smoothed image; an absolute-value acquiring process for acquiring an absolute value image from a primary differential image; a maximum-value acquiring process relative to each absolute value image in the tree directions; a binarization process relative to a maximum value image; a labeling process relative to a binary image; a domain extraction process for extracting a label A domain; an irregularity value operation process for operating an irregularity value in the label A domain; and a quality determination process for determining the quality of the label A domain based on the irregularity value. This device to which the method is applied is also provided. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention belongs to the technical field of inspecting whether or not optical characteristics of an object are uniform. In particular, the present invention relates to a method and apparatus for inspecting streaky uneven defects that can obtain high inspection performance without being affected by the width in the short side direction of streaky uneven defects.

In many conventional inspection apparatuses for inspecting streaky unevenness defects, a secondary differential filter is applied in the process of extracting streaky unevenness defects from an input image obtained by imaging an object (Patent Document 1). Patent Document 2). In order to accurately extract streak-like unevenness using a second-order differential filter, it is necessary to apply a second-order differential filter that matches the width of the streaky unevenness defect in the short side direction, as shown in FIG. . However, it is impossible to know the width in the short side direction generated in the object before the inspection. In addition, there are limits to the types of secondary differential filters that can be prepared and applied in advance, and they are not necessarily compatible. Therefore, sufficient inspection performance could not be obtained.
JP 2000-199745 A JP 2002-28059 A

  The present invention has been made to solve the above problems. An object of the present invention is to provide a method and apparatus for inspecting streaky unevenness defects that can obtain high inspection performance without being affected by the width in the short side direction of streaky unevenness defects.

According to a first aspect of the present invention, there is provided a method for inspecting a stripe-like unevenness defect, wherein a noise removal process for obtaining a noise-removed image by applying a noise removal filter to an input image, A smoothing process for obtaining a smoothed image by applying a smoothing filter in the direction of the unevenness defect, and the streak-like unevenness to be detected from four directions of the vertical direction, the horizontal direction, and the two diagonal directions with respect to the smoothed image. Applying a first-order differential filter to the three directions excluding the defect direction to obtain a first-order differential image in the three directions, and the pixel values of the pixels of the first-order differential image in the three directions are set to absolute values thereof. The maximum value image in which the absolute value process in which the absolute value image in the three directions is replaced and the pixel value of the pixel at the same position in the absolute value image in the three directions is compared and the largest pixel value is the pixel value of the pixel Maximization process to get A binarization process for obtaining a binary image by applying a predetermined threshold to the maximum value image, a labeling process for labeling the binary image to obtain a labeling image, and a labeling number from the labeling image. An area extraction process for extracting the label A area as A, a mura value calculation process for calculating a mura value that is a difference between pixel values of the mura area and the non-mura area in the label A area, and the mura value based on the mura value. A pass / fail judgment process for judging pass / fail of the label A area.
According to a second aspect of the present invention, there is provided a method for inspecting a stripe-like unevenness defect, wherein the unevenness value calculating step includes a step of detecting a pixel B constituting a boundary of the label A region. A boundary extraction process for extracting and obtaining a boundary image, a coordinate extraction process for extracting the coordinates (X [B], Y [B]) of the pixel B that constitutes the boundary, and the absolute values in the four directions by paying attention to the coordinates An image extraction process for extracting the absolute value image having the largest pixel value of the pixel at the coordinates from the image, and a change direction extraction process for setting the direction of the primary differential filter applied to the extracted absolute value image as the change direction of the pixel value Extracting a pixel coordinate from the coordinate to the boundary of the label A region in the change direction to obtain a cross region coordinate, and a maximum of pixels of the cross region coordinate in the noise-removed image A pixel value difference calculation process for calculating a pixel value difference, which is a difference between an elementary value and a minimum pixel value, and an average value of the pixel value differences obtained for all the pixels B constituting the boundary are calculated, and the average value is calculated. And an average value calculation process for making the unevenness value.
According to a third aspect of the present invention, there is provided a streak-like unevenness defect inspection apparatus comprising a line sensor camera, a conveying means, and a processing means, wherein the line sensor camera has a linear shape. The main scanning of the imaging region is performed to image the inspection target article and an imaging signal is output. The transport unit transports the inspection target article in the sub-scanning direction with respect to the main scanning. The imaging signal is input in synchronization with sub-scanning to generate an input image, and the input image is subjected to data processing to which the method for inspecting streaky unevenness defects according to claim 1 is applied. The quality is determined.

According to the method for inspecting streaky unevenness according to claim 1 of the present invention, a noise removal filter is applied to an input image in a noise removal process to obtain a noise removal image, and a noise removal image is obtained in a smoothing process. A smoothing filter in the direction of the stripe-shaped unevenness to be detected is applied to obtain a smoothed image, and in the first-order differentiation process, the smoothed image is selected from four directions of the vertical direction, the horizontal direction, and the two diagonal directions. A first-order differential image is obtained by applying a first-order differential filter to the three directions excluding the direction of the stripe-shaped unevenness to be detected, and the pixel values of the pixels of the first-order differential image in the three directions are obtained in the absolute value conversion process. Is replaced with the absolute value to obtain an absolute value image in three directions, and in the maximization process, the pixel values of the pixels at the same position in the absolute value image in three directions are compared to determine the largest pixel value. A maximum value image is obtained, a predetermined threshold is applied to the maximum value image in the binarization process, and a binary image is obtained. In the labeling process, labeling is performed on the binary image to obtain a labeling image. In the area extraction process, the label A area with the label ring number A is extracted from the labeling image, and the unevenness value, which is the difference between the pixel values of the uneven area and the non-mura area in the label A area, is calculated in the unevenness value calculation process. In the quality determination process, the quality of the label A area is determined based on the unevenness value. That is, the primary differential filter is applied without applying the secondary differential filter that is affected by the width in the short side direction of the stripe-like unevenness defect. In the conventional secondary differentiation process, the uneven area itself is detected, whereas in the primary differential process, the boundary (pixel value (luminance) change area) between the uneven area and the non-uniform area is extracted. Unevenness can be extracted regardless of the width (see FIG. 8). Therefore, there is provided a method for inspecting a stripe-like unevenness defect that can obtain a high inspection performance without being affected by the width in the short side direction of the stripe-like unevenness defect.
According to the method for inspecting streaky mura defect according to claim 2 of the present invention, the mura value calculation process includes a boundary extraction process, a coordinate extraction process, an image extraction process, a change direction extraction process, an intersection area extraction process, and a pixel value difference. A calculation process and an average value calculation process; in the boundary extraction process, a pixel B constituting the boundary of the label A region is extracted to obtain a boundary image; and in the coordinate extraction process, the coordinates (X [ B], Y [B]) are extracted, paying attention to the coordinates in the image extraction process, the absolute value image having the largest pixel value of the coordinate pixel is extracted from the absolute value images in the four directions, and extracted in the change direction extraction process The direction of the primary differential filter applied to the absolute value image is the change direction of the pixel value, and in the intersection region extraction process, the coordinates of the pixel from the coordinate to the boundary of the label A region in the change direction are extracted. All pixel values are calculated as the difference area coordinates, and the pixel value difference, which is the difference between the maximum pixel value and the minimum pixel value of the intersection area coordinates in the noise-removed image, is calculated in the pixel value difference calculation process. The average value of the pixel value differences obtained for the pixel B is calculated, and the average value is set as the uneven value. That is, a pixel value change direction is searched from the extracted boundary image, and a mura value, which is a difference between the mura region and the non-mura region, is obtained as a difference between the maximum pixel value and the minimum pixel value in the direction. Therefore, the unevenness value can be calculated with high accuracy.
According to the striped uneven defect inspection apparatus of the third aspect of the present invention, the main scanning of the linear imaging region is performed by the line sensor camera, the inspection target article is imaged, and the imaging signal is output, and the conveying means is used. The article to be inspected is conveyed in the direction of sub-scanning with respect to main scanning, and an imaging signal is input by the processing means in synchronization with main scanning and sub-scanning to generate an input image. Data processing to which the inspection method for streaky unevenness defects is applied is performed, and the quality of the inspection target article is determined. That is, the primary differential filter is applied without applying the secondary differential filter that is affected by the width in the short side direction of the stripe-like unevenness defect. In the conventional secondary differentiation process, the uneven area itself is detected, whereas in the primary differential process, the boundary (pixel value (luminance) change area) between the uneven area and the non-uniform area is extracted. Unevenness can be extracted regardless of the width (see FIG. 8). Therefore, an inspection apparatus for streaky unevenness defects that can obtain high inspection performance without being affected by the width in the short side direction of streaky unevenness defects is provided.

Next, embodiments of the present invention will be described with reference to the drawings. An example of the configuration of the inspection device for streaky unevenness defects of the present invention is shown in FIG. In FIG. 1, 1 is a line sensor camera, 2 is a light source, 3 is an image processing unit, 4 is an input unit, 5 is an output unit, 6 is a conveyor, and 100 is an object.
The object 100 may be any article and is not particularly limited. For example, it is an article having a surface to be inspected, such as a web, a sheet, and a panel. The traveling direction of the object 100 is a direction indicated by an arrow in FIG.

  The line sensor camera 1 includes a line sensor element, an output amplifier, a drive circuit for outputting signals in time series, an imaging lens, and the like, and has a linear imaging region. The line sensor element is a large scale integrated circuit (LSI) such as a charge coupled device (CCD) or a metal oxide semiconductor (MOS) in which a plurality of light receiving portions are arranged in a straight line. As shown in FIG. 1, the imaging area of the line sensor camera 1 extends in the width direction of the object 100 (a method perpendicular to the traveling direction of the object 100). Imaging of the surface (two-dimensional area) of the object 100 can be performed by main scanning by the line sensor camera 1 and sub-scanning by the object 100 traveling. The transporter 6 is a transporter that transports the object 100 in the sub-scanning direction.

  In the example shown in FIG. 1, the optical axis of the line sensor camera 1 (center line in the imaging direction) deviates from the vertical direction with respect to the surface of the object 100 and has a predetermined angle. That is, the line sensor camera 1 images the surface of the object 100 from an oblique direction. The angle formed by the optical axis of the line sensor camera 1 and the surface of the object 100 is an appropriate angle depending on the characteristics of the defect to be imaged. The angle is set such that light modulation peculiar to the defect can be imaged according to the characteristic of the defect.

  The light source 2 is an illumination unit that illuminates the imaging region of the line sensor camera 1. The light source 2 is preferably a light source capable of obtaining a linear illumination area so as to match the linear imaging area of the line sensor camera 1. In general, the illumination area by the light source 2 is often set so as to include the imaging area of the line sensor camera 1. In order to obtain a linear illumination region, a light source that emits linear light is used as the light source 2. As a light source that emits light in a straight line, for example, a straight tube fluorescent lamp, a straight tube halogen lamp, a light source in which LEDs (light emitting diodes) are arranged in a straight line, a light beam of a point light source is guided by an optical fiber, and is linear. An optical fiber light source adapted to irradiate a light source, a light source adapted to direct light from a point light source through a light guide tube and irradiate from a slit, and the like can be used.

  In FIG. 1, a light source 2 that emits light in a straight line is used as a light source of a reflective imaging system, and the light source 2 is disposed together with the line sensor camera 1 on the surface side of an object 100. The direction in which the light source 2 that performs linear light emission extends and the direction in which the linear imaging region extends are parallel to each other. The relationship between the optical axis of the line sensor camera 1 and the arrangement of the light sources 2 is determined from comprehensive judgment in consideration of facilitating imaging of defects as described above in the line sensor camera 1.

  The image processing unit 3 includes an image storage unit that inputs an imaging signal output from the line sensor camera 1 and stores it as an input image, that is, an image memory. The image processing unit 3 performs image data processing such as extracting streak-like unevenness defects for the image stored in the image memory, data processing related to settings and operations related to the image input device, data processing related to the user interface, etc. Data processing is performed. In connection with the user interface, an input unit 4 such as a keyboard and a mouse, and an output unit 5 such as a display and an alarm device are connected to the image processing unit 3.

  The configuration of the streaky unevenness defect inspection apparatus according to the present invention has been described above. Next, operations in the method and apparatus for inspecting streaky unevenness of the present invention, that is, operations in the image processing unit 3 will be described with reference to the drawings. FIG. 2 is a schematic diagram showing the overall operation of the image processing unit 3 of the present invention. FIG. 3 shows an example of a data processing process of inspection for extracting streaky unevenness defects from an input image and judging whether or not the product is acceptable. FIG. 4 shows an example of a smoothing filter used in inspection data processing, and FIG. 5 shows an example of a primary differential filter.

  In the present invention, as shown in FIG. 2, based on the noise-removed image obtained by performing the noise removal process, a process for detecting vertical stripe-like unevenness defects and a horizontal stripe-like unevenness defect are detected. A total of four processes including a process, a process for detecting a stripe-shaped uneven defect in a 45-degree oblique direction, and a process for detecting a stripe-shaped uneven defect in a 135-degree oblique direction are applied to detect a stripe-shaped uneven defect in all directions. In the flowchart shown in FIG. 3, only the process for detecting the vertical stripe-like unevenness defect therein is described. The other three processes can be easily inferred from the description with reference to the flowchart, and the description thereof will be omitted.

  First, in step S1 (noise removal process) in FIG. 3, a noise removal filter is applied to the input image input to the image memory to perform noise removal processing to obtain a noise removal image. As a noise removal filter, there are a Gaussian filter, a median filter, and an average value filter, and noise removal is performed by applying a filter suitable for the noise to be removed. For example, a median filter is applied to salt and pepper noise, a Gaussian filter is applied to weak noise, and an average filter is applied to strong noise.

  Next, in step S2 (smoothing process), a smoothing image is obtained by applying a smoothing filter in the direction of streaky unevenness to be detected to the noise-removed image. In this case (in the flowchart shown in FIG. 3), as an example, the direction of the stripe-shaped unevenness to be detected is assumed to be the vertical direction, and the horizontal direction, the vertical direction, the 45 ° oblique direction, and the oblique direction shown in FIG. A vertical direction smoothing filter is applied from the 135 degree direction.

  Next, in step S3 (first-order differentiation process), a first-order differential filter such as a Sobel filter or a Prewitt filter is applied to the noise-removed image, and a first-order differential process is performed to obtain a first-order differential image. Since the change in pixel values of the non-uniformity region which is not the region of the stripe-like unevenness and the unevenness region which is the stripe-like unevenness defect is generally gentle, the positive coefficient pixel and the negative coefficient in the first-order differential filter 3 pixels or more are separated from each other. As for the direction of the first-order differential filter, the first-order differential filter in three directions excluding the direction of the stripe-shaped unevenness to be detected from the four directions of the vertical direction, the horizontal direction, and the two diagonal directions with respect to the smoothed image. Is applied in three directions. That is, the horizontal direction and the diagonal two directions are set to three directions, and the first-order differential filter in the three directions is applied.

  Therefore, in step S3, each of the three-direction primary differential filters is applied to the smoothed image to perform the primary differential processing, and three (three) primary differential images are obtained. FIG. 5A shows a first-order differential filter in the horizontal direction, FIG. 5B shows a first-order differential filter in the vertical direction, and FIG. 5C shows an oblique 45-degree direction. FIG. 6D shows a first-order differential filter in the oblique 135 degree direction.

Next, in step S4 (absolute value conversion process), for each of the first-order differential images in the three directions, absolute value processing is performed to replace the pixel values of the pixels constituting the image with the absolute values of the pixel values. Obtain an absolute value image.
Next, in step S5 (maximization process), maximum value processing is performed based on the absolute value images in three directions to obtain a maximum value image. The maximum value processing is processing in which pixel values of pixels at corresponding positions in a plurality of images are compared and the maximum pixel value is set as a pixel value of a pixel at a corresponding position in the maximum value image.

Next, in step S6 (binarization process), a binarization process is performed on the maximum value image to obtain a binarized image. The binarization threshold value is set in advance so that the boundary between the non-uniformity region and the non-uniformity region to be detected becomes the pixel value 1 in the binary image.
Next, in step S7 (labeling process), a labeling image is obtained by performing a labeling process on the binarized image.
Next, in step S8 (region extraction process), region extraction processing for extracting a label A region with a labeling number A from the labeling image is performed.

Next, in step S9 (unevenness value calculation process), unevenness value calculation processing for calculating an unevenness value which is a difference between pixel values of the unevenness area and the non-nonuniformity area in the label A area is performed.
Next, in step S10 (quality determination process), quality determination processing for determining quality of the label A area based on the unevenness value is performed.
Then, when all the labeling numbers in the labeling image have not been judged as good or bad, the labeling number A is changed to a labeling number that has not been judged good or bad, and the above-described steps after step S7 are repeated. When the pass / fail determination is completed for all the labeling numbers in the labeling image, the data related to the above-described processing (pass / fail determination, intermediate image, log, processing condition), particularly the pass / fail determination data is stored in the memory.

The image processing in the method and apparatus for inspecting streaky unevenness defects of the present invention has been described above with reference to the drawings. Next, step S9 (uneven value calculation process) will be described in detail with reference to the drawings. FIG. 6 shows the details of an example of the unevenness value calculation process in the method and apparatus for inspecting streaky unevenness defects of the present invention.
First, in step S8 (region extraction process) of FIGS. 4 and 6, the label A region is extracted.

  Next, in step S101 (boundary extraction process) in FIG. 6, pixels constituting the boundary of the label A region are extracted to obtain a boundary image. Note that the boundary is not a pixel because it determines whether or not a specific pixel is included in a specific region (set). However, here, a pixel adjacent to the boundary is referred to as a pixel constituting the boundary. The pixels constituting the boundary include a pixel existing inside a region determined by the boundary and a pixel existing outside. In the present invention, even if the pixels constituting the boundary are defined either inside or outside, usually, the determination is not greatly affected.

Next, in step S102 (coordinate extraction process), pixel coordinates (X [i], Y [i]) in the boundary image are extracted. However, i = 1 to n (n: total number).
Next, in step S103 (initialization process), i = 1 is set.
Next, in step S104 (focus process), the pixel Pi at the coordinates (X [i], Y [i]) is set as a processing target (focussed).
Next, in step S105 (image extraction process), the absolute value having the largest pixel value (luminance) of the pixel Pi at coordinates (X [i], Y [i]) from the absolute value image in four directions (see step S3). Extract images.

Next, in step S106 (change direction extraction process), change direction extraction processing is performed in which the direction of the primary differential filter applied to the extracted absolute value image is the change direction of the pixel value (luminance).
Next, in step S107 (straight line drawing process), a straight line that passes through the pixel Pi at the coordinates (X [i], Y [i]) and extends in the change direction of the pixel value is applied (drawn) to the label A region.
Next, in step S108 (intersection area extraction process), an area where the fitted straight line and the pixel in the label A area intersect is extracted. That is, the intersection area extraction process is performed by extracting the coordinates of the pixels from the pixel Pi at the coordinates (X [i], Y [i]) until the intersection with the boundary of the label A area in the change direction to obtain the intersection area coordinates.

Next, in step S109 (pixel value difference calculation process), a pixel value difference calculation that calculates a pixel value difference that is the difference between the maximum pixel value and the minimum pixel value of the pixels in the intersection region coordinates in the noise-removed image (see step S1). Process.
Next, in step S110 (total number?), It is determined whether or not the pixel value difference has been calculated for all the pixels constituting the boundary. That is, it is determined whether i = N. When i = N, the process proceeds to step S111. When i is not N, the process proceeds to step S112.

In step S111 (average value calculation process), an average value of pixel value differences obtained for all the pixels Pi (i = 1 to n) constituting the boundary is calculated, and an average value calculation using the average value as an uneven value. Process.
In step S112 (continuation process), i = i + 1 is set, the process returns to step S104, and the subsequent steps are repeated.

  As described above, in the present invention, the primary differential filter is applied without applying the secondary differential filter affected by the width in the short side direction of the stripe-like unevenness defect. In the secondary differentiation process, the uneven area itself is detected, whereas in the primary differential process, the boundary between the uneven area and the non-uniform area is extracted, so that the unevenness can be extracted regardless of the width in the short side direction. . In addition, the pixel value change direction is searched from the extracted boundary image, and the unevenness value, which is the difference between the unevenness region and the non-nonuniformity region, is calculated as the difference between the maximum pixel value and the minimum pixel value in that direction. Is highly accurate. Accordingly, there is provided an inspection method and apparatus for streaky unevenness defects that can obtain high inspection performance without being affected by the width in the short side direction of streaky unevenness defects.

It is a figure which shows an example of the structure in the inspection apparatus of the stripe-shaped nonuniformity defect of this invention. It is the schematic which shows the whole operation | movement in the image processing part of this invention. It is a flowchart which shows an example of the data processing process of the test | inspection which extracts a stripe-shaped nonuniformity defect from an input image and determines a quality. It is a figure which shows an example of the smoothing filter used in the data processing of a test | inspection. It is a figure which shows an example of the primary differential filter used in the data processing of a test | inspection. It is a figure which shows the detail about an example of the process of calculating the brightness | luminance difference in the data processing of a test | inspection. It is explanatory drawing which shows applying the secondary differential filter which adapts to the width | variety of the short side direction of a stripe-shaped nonuniformity defect. It is explanatory drawing which shows the difference of a primary differentiation process and a secondary differentiation process.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Line sensor camera 2 Light source 3 Image processing part 4 Input part 5 Output part 6 Conveyor 100 Target object

Claims (3)

Applying a noise removal filter to the input image to obtain a noise removal image,
Applying a smoothing filter in the direction of streaky unevenness to be detected to the noise-removed image to obtain a smoothed image; and
A first-order differential filter is applied to three directions excluding the direction of the stripe-like unevenness to be detected from four directions of the vertical direction, the horizontal direction, and the two diagonal directions with respect to the smoothed image. A first derivative process for obtaining a second derivative image;
An absolute value conversion process of obtaining the absolute value image in the three directions by replacing the pixel value of the pixel of the primary differential image in the three directions with the absolute value;
A maximization process in which pixel values of pixels at the same position in the absolute value image in the three directions are compared to obtain a maximum value image having the largest pixel value as the pixel value of the pixel;
A binarization process for obtaining a binary image by applying a predetermined threshold to the maximum value image;
A labeling process of labeling the binary image to obtain a labeled image;
An area extraction process for extracting a label A area having a labeling number A from the labeling image;
A mura value calculation process for calculating a mura value, which is a difference between pixel values of the mura area and the non-mura area in the label A area,
A pass / fail determination process for determining pass / fail of the label A area based on the unevenness value;
A method for inspecting streaky unevenness defects, comprising: The method for inspecting streaky unevenness defects according to claim 1, wherein the unevenness value calculating step is:
A boundary extraction process for extracting a pixel B constituting a boundary of the label A region to obtain a boundary image;
A coordinate extraction process for extracting the coordinates (X [B], Y [B]) of the pixel B constituting the boundary;
An image extraction process of focusing on the coordinates and extracting an absolute value image having the largest pixel value of the pixels of the coordinates from the absolute value images in the four directions;
A change direction extraction process in which the direction of the primary differential filter applied to the extracted absolute value image is the change direction of the pixel value;
An intersection region extraction process in which the coordinates of the pixels from the coordinates to the intersection of the label A region in the change direction are extracted and used as intersection region coordinates;
A pixel value difference calculation process for calculating a pixel value difference that is a difference between the maximum pixel value and the minimum pixel value of the pixels of the intersection region coordinates in the noise-removed image;
An average value calculation process of calculating an average value of the pixel value differences obtained for all the pixels B constituting the boundary and setting the average value as the unevenness value;
A method for inspecting streaky unevenness defects, comprising: An inspection apparatus for streaky uneven defects comprising a line sensor camera, a conveying means, and a processing means,
The line sensor camera performs main scanning of a linear imaging region, images an inspection target article, and outputs an imaging signal;
The conveying means conveys the inspection target article in a sub-scanning direction with respect to the main scanning;
The processing means inputs the imaging signal in synchronization with the main scanning and the sub-scanning to generate an input image, and applies the method for inspecting streaky unevenness defects to the input image. An inspection apparatus for streaky unevenness defects, wherein data processing is performed to determine whether the inspection target article is good or bad.