JP2016109437A - Image processing device and method, and defect inspection device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect a defect with high accuracy without having a plurality of light sources and means for controlling the light sources in a defect inspection device for detecting a defect such as a flaw or scratch on the surface of a sheet product.SOLUTION: Provided is an image processing device equipped with image processing means and defect detection means for detecting a defect on the basis of an image to which image processing has been applied, wherein the image processing means includes address arithmetic means for calculating an address for accessing pixel data stored for image processing, pixel data access means for reading or writing the pixel data of image storage means on the basis of the calculated address, pixel data storage means for temporarily storing read or written pixel data, and pixel data arithmetic processing means for rewriting temporarily stored pixel data with the calculation result of the pixel data by image processing, the image processing being performed on the pixel data of peripheral pixels separate by a prescribed number of pixels in accordance with the sizes of a pixel of interest that is the object of image processing and a defect assumed from the pixel of interest.SELECTED DRAWING: Figure 2

Description

  The present invention relates to an image processing apparatus and method, and a defect inspection apparatus.

  2. Description of the Related Art In a production process of a sheet-like sheet product such as paper and film (hereinafter referred to as a sheet product), there is a defect inspection apparatus that inspects defects such as scratches and rubbing on the surface using the product as an inspection object. As an example, using the fact that there is a difference in density between the defective part and the part that is not, performing the differentiation process of the gray image obtained by imaging the object to be inspected with the light from the light source with a camera, A technique for detecting, as a defect, a portion having an edge intensity greater than a predetermined threshold value from an edge-enhanced image obtained thereby is already known.

  In Patent Document 1, a state of a light source and the like based on image data obtained by actually capturing an object to be inspected in order to obtain an optimal image suitable for an inspection purpose for realizing a high-precision inspection is disclosed. A configuration for automatically or semi-automatically determining or selecting conditions is disclosed.

  However, in the defect inspection apparatus according to the prior art, in order to be able to distinguish from the difference in density due to the roughness on the surface, the difference in density between the defective part and the other part needs to be relatively large. In the case of sheet-like products such as paper and film, defects such as scratches and rubbing on the surface of the product are slight differences in gradation that can be slightly recognized by human eyes, and changes in the gradation are also gradual. Since the pattern differs depending on the defect, it is difficult to detect only the defect because it cannot be distinguished from the density difference due to the roughness on the surface. Therefore, it is necessary to adjust the state of the light source brightness, light wavelength, irradiation position, etc. according to the detected defect so that the above distinction can be made. However, there is a problem that the optical system is complicated and the apparatus becomes large. In particular, in Patent Document 1, the problem that the optical system is complicated and the apparatus becomes large cannot be solved.

  An object of the present invention is to provide a defect inspection apparatus for detecting defects such as scratches and rubbing on the surface of a sheet-like sheet product such as paper and film. An object of the present invention is to provide an image processing apparatus capable of detecting the above.

An image processing apparatus according to an aspect of the present invention includes:
Image storage means for storing an image obtained by imaging the inspection object;
Image processing means for performing predetermined image processing on the image stored in the image storage means;
In an image processing apparatus provided with defect detection means for detecting defects based on the image subjected to the image processing,
The image processing means includes
Address calculating means for calculating an address for accessing pixel data stored in the image storage means for the image processing;
Pixel data access means for reading or writing pixel data of the image storage means based on the calculated address;
Pixel data storage means for temporarily storing the pixel data to be read or written;
Pixel data calculation processing means for rewriting the temporarily stored pixel data with the calculation result of the pixel data by the image processing;
Image processing is performed on pixel data of peripheral pixels separated by a predetermined number of pixels in accordance with the target pixel to be processed and the size of a defect assumed from the target pixel.

  Therefore, according to the image processing apparatus of the present invention, defects can be detected with high accuracy without increasing the size of the apparatus.

It is a block diagram which shows the structure of the defect inspection system which concerns on one Embodiment of this invention. It is a block diagram which shows the structure of the image processing apparatus 3 of FIG. It is a flowchart which shows the 1st part of the image result detection process performed by the image processing apparatus 3 of FIG. It is a flowchart which shows the 2nd part of the image result detection process performed by the image processing apparatus 3 of FIG. It is a 1st example of the image inspected by the defect inspection system concerning a prior art, Comprising: It is a top view which extracted a part of the image, and one ridge shows one pixel. It is an example of the image test | inspected by the defect inspection system of FIG. 1, Comprising: It is the top view which extracted a part of the said image, and one ridge shows one pixel. It is a graph when the change of the pixel value around the defect in the said image is cut out in one direction with respect to an example of the image inspected by the defect inspection system of FIG. It is the front view of an example of the image of the to-be-inspected object with the defect 101, and the enlarged view which extracted and illustrated the part 102 which is the part. It is a front view which shows the pixel value which represented the part of the enlarged view of FIG. 8 by image data. It is a front view which shows the result of having processed the image data of FIG. 9 by the edge detection process which concerns on a prior art. It is a front view which shows the result of having processed the image data of FIG. 9 by the edge detection process which concerns on this embodiment. It is a front view of an image showing image processing when pixel data for performing image processing is a characteristic value of pixel data within a predetermined vertical and horizontal rectangular range centered on each pixel of peripheral pixels. It is a front view of the image which shows image processing when each rectangular range centering on a surrounding pixel differs in aspect ratio by the shape of the defect assumed beforehand.

  Hereinafter, embodiments according to the present invention will be described with reference to the drawings. In addition, in each following embodiment, the same code | symbol is attached | subjected about the same component.

  FIG. 1 is a block diagram showing a configuration of a defect inspection system according to an embodiment of the present invention. The defect inspection system according to the present embodiment has the following characteristics in detection processing for detecting defects such as scratches and rubbing on the surface of a sheet-like sheet product such as paper or film. In other words, the difference between the defective part and the non-existent part is very small, and the change in the density is also moderate. Therefore, the edge strength obtained is also small. Therefore, it is utilized that the edge strength obtained by increasing the difference between the two pixel values by performing differentiation using the pixel of interest and an arbitrary distance from the position is increased. A defect is detected by performing image processing to be described later in detail.

  In FIG. 1, a sheet-like sheet product 1 such as paper or film, which is an object to be inspected, is conveyed at a constant speed in the direction of arrow A1 using a conveying device such as ADF (Automatic Document Feeder). At this time, the camera 2 captures the surface of the sheet product 1 at regular intervals in synchronization with the flow of the sheet product 1, and sends the image data to the image processing device 3. The image processing device 3 performs image processing for each image data sent from the camera 2 to perform defect inspection, and displays the inspection result on the display device 4 such as a liquid crystal display.

  FIG. 2 is a block diagram showing the configuration of the image processing apparatus 3 of FIG. In FIG. 2, the image processing apparatus 3 includes an image input unit 11, an image storage unit 12, a defect detection unit 13, and an image processing unit 10. Here, the image processing unit 10 includes an address calculation unit 14, a pixel data access unit 15, a pixel data storage unit 16, and a pixel data calculation processing unit 17. Here, the image processing device 3 performs predetermined image processing on the image data received from the camera 2, detects a defect, and outputs the result to the display device 4.

  In FIG. 2, the image input unit 11 records the received image data in the image storage unit 12. The image processing unit 10 reads the image data recorded in the image storage unit 12 for each pixel, and records the processed image data in another free area of the image storage unit 12 as processed image data. The image storage unit 12 is configured by a memory medium such as a DRAM or SRAM, or a storage medium such as a hard disk, and a configuration in which the storage medium is used alone or a combination of a plurality of them is also considered. It is done.

  The defect detection unit 13 detects defects from the processed image data. For example, the value for each pixel in the processed image data is checked, and if it is less than a predetermined value, it is judged as non-defective, and if it is above, it is judged as defective, and the information is output to the display device 4.

  The image processing unit 10 in FIG. 2 performs predetermined image processing, which will be described in detail later, for each pixel on the image data stored in the image storage unit 12. The address calculation unit 14 calculates an address for reading the pixel to be processed and its surrounding pixels from the image storage unit 12 and an address for writing back the processed pixel to the image storage unit 12 and an address of the calculation result Is output to the pixel data access unit 15. The pixel data access unit 15 reads pixel data from the image storage unit 12 based on the address calculated by the address calculation unit, temporarily stores it in the pixel data storage unit 16, or conversely, after processing from the pixel data storage unit 16 Pixel data is read and written in the image storage unit 12.

  The pixel data storage unit 16 stores the pixel data of the target pixel, which is the image processing target pixel written by the pixel data access unit 15, and the pixel data of a plurality of peripheral pixels around the pixel necessary for the processing. It also stores data. The pixel data calculation processing unit 17 reads the data of the target pixel and the plurality of peripheral pixels in the pixel data storage unit 16 and records the processed pixel data in the pixel data storage unit 16 after performing, for example, an edge detection processing calculation.

  The peripheral pixels are pixels that are a predetermined number of pixels away from the pixel of interest. For example, in the XY two-dimensional coordinates, the pixel coordinates to be processed are (x, y), and the predetermined number of pixels is w. At this time, the coordinates of the peripheral pixels are (x−w, y−w), (x, y−w), (x + w, y−w), (x−w, y), (x + w, y), (x −w, y + w), (x, y + w), and (x + w, y + w).

  3 and 4 are flowcharts showing an image result detection process executed by the image processing apparatus 3 of FIG.

  In FIG. 3, first, the image data captured from the image input unit 11 in step S <b> 1 is stored in the image storage unit 12. Next, in step S2, the calculation result output destination address ADDrGout is substituted into a variable n and set. That is, the head address of the empty area on the image storage unit 12 for storing the data after image processing is set, and the same value is set as an initial value for the variable n for sequentially storing the data of the processing result.

  Next, in step S3, the vertical coordinate value x of the image is initialized to 0, and in step S4, the horizontal coordinate value y of the image is initialized to 0. The double loop from step S5 to step S13 in FIG. 3 is for scanning the horizontal coordinate value x and the vertical coordinate value y of the image, and the pixel at the coordinate (x, y) on the image is the processing target. The pixel of interest is a pixel. The pixel of interest starts from the upper left pixel on the image and moves to the right by one pixel. When it reaches the right end of the image, it moves to the left end pixel one pixel below. From there, it is repeated until it reaches the lower right pixel of the image (steps S10 to S13).

  In step S5, the coordinates of the eight neighboring pixels described above from the coordinates (x, y) indicated at that time and the addresses AddrG1 to AddrG8 on the image storage unit 12 corresponding to the respective coordinates are calculated. . Next, in step S6, the pixel data of these eight addresses AddrG1 to AddrG8 are read from the image storage unit 12 and stored in the arrays G [1] to G [8] in this example. Further, in step S7, an edge detection calculation is performed using the eight pixel data prepared in the array, and the result is stored in the array G [0] in step S8. The calculation method of the edge detection calculation will be described later in detail. The data storage destination may be an individual variable or register instead of an array.

  Next, in step S9, G [0] data is stored at address n where the processing result data is stored, and address n is incremented by 1 in this example for the next storage destination. Depending on the system used, the value added to n varies depending on the configuration of the system used, such as adding 2 when the data width extends over a plurality of addresses, for example, when extending over 2 addresses.

  Here, the processing from step S5 to S9 is performed for each pixel, and the processing of all the pixels of the image is completed when the previous double loop is completed. And if it is YES at step S12, it will progress to step S21 of FIG.

  In step S21 of FIG. 4, the output destination address ADDrGout is substituted for the variable N. The loop processing from step S22 to step S26 in FIG. 4 is for checking the processing result data, and the address from the beginning to the end of the processing result data is substituted for n (steps S22 to S27).

  In step S22, the data indicated by the address n is read from the image storage unit 12 and substituted for the variable d. Next, in step S23, the value of the variable d is compared with a predetermined threshold value. If the variable d is smaller than the threshold value, it is determined as non-defective in step S24. It is determined that the defect information is output. Note that the threshold value is set in advance to a value at which it can be detected according to a defect assumed in the inspection object. In step S25, the data determined as a defect, the data value, the coordinates on the original image calculated from the address n, and the like are output. The output destination is the display device 4 such as a monitor or a device for notifying the detection of a defect as required by the inspection system.

  When the loop processing from step S22 to step S27 is completed, the inspection of one image data is completed. In step S28, it is determined whether there is a next inspection target image. If there is, the process returns to step S1 in FIG. 4 to repeat the process, and if not, the process ends.

  FIG. 5 is a first example of an image to be inspected by the defect inspection system according to the related art, and is a plan view in which a part of the image is extracted, and one ridge indicates one pixel. A conventional edge detection process will be described with reference to FIG.

  In FIG. 5, when the pixel G0 is a pixel to be processed, the edge detection processing according to the conventional technique performs an operation based on the pixel data of the target pixel and the eight adjacent pixels G1 to G8 adjacent thereto, and calculates the edge strength as the operation result. obtain. Several kinds of calculation methods for edge detection are known. Basically, the absolute value of the gradient of the luminance change, which is the magnitude of the change in the pixel value, is used as the edge intensity. A certain Sobel operator performs the following calculation. Here, assuming that the pixels G0 to G8 are pixel data of each pixel, the edge detection calculation result f is expressed by the following equation.

fx = − (G1 + (2 × G4) + G6) + (G3 + (2 × G5) + G8)
fy = − (G1 + (2 × G2) + G3) + (G6 + (2 × G7) + G8)

  The calculation result f becomes the edge intensity of the pixel G0, and this value is larger in a place where there is a defect than in a place where there is no defect, so that the defect can be detected. In the image data obtained by imaging the object to be inspected, the luminance of the defective portion is lower or higher than the surrounding area. Although this varies depending on the type of defect, since the edge strength is to obtain the absolute value of the gradient of the luminance change, there is no distinction between the two.

  FIG. 6 is an example of an image inspected by the defect inspection system of FIG. 1, and is a plan view of a part of the image, where one ridge indicates one pixel. With reference to FIG. 6, a specific edge detection process according to the present embodiment will be described.

  In FIG. 6, assuming that the pixel G0 is a pixel of interest, a calculation is performed based on the pixel values of eight peripheral pixels G1 to G8 that are separated from the pixel of interest by an arbitrary number of pixels, and edge strength is obtained as a calculation result. The calculation method is the same as the conventional edge detection calculation method. Since the edge strength is the absolute value of the slope of the luminance change, if the pixel value difference between the pixels is the same, the slope decreases as the distance increases, but since the distance between the pixels is not considered in the calculation, the edge strength is This is proportional to the difference in pixel value between pixels.

  FIG. 7 is a graph when a change in pixel values around the defect in the image is cut out in one direction with respect to an example of the image inspected by the defect inspection system in FIG. With reference to FIG. 7, a specific edge detection process according to the present embodiment will be described.

  In FIG. 7, the difference between the pixel value of the non-defective part and the pixel value of the defective part is small at the end and is maximum at the center. Since the edge strength is proportional to the difference in pixel value between pixels, the maximum value is the predetermined unit width so that the pixel G0 in FIG. 6 is the target pixel and the pixel value includes the maximum and minimum pixels. This is a case where a pixel separated by w1 is a peripheral pixel. This means that when the width of the defective portion is W, the unit width w1 may be reduced to 1/4 of the width. Since defects occurring in the actual manufacturing process are predicted in advance, the arbitrary number of pixels described in FIG. 5 may be an integer around 1/4 of the width of the predicted defect.

  FIG. 8 is a front view of an example of an image of an inspection object having a defect 101, and an enlarged view showing a portion 102 which is a part of the image. A specific edge detection process according to the present embodiment will be described with reference to FIG.

  In FIG. 8, one ridge in the enlarged view represents one pixel, and the hatched portion represents a defective pixel. In the edge detection process, for example, the target pixel is started from the upper left pixel of the image of the object to be inspected as indicated by an arrow A2 in the figure, and the process is sequentially performed while shifting one pixel at a time in the right direction. When it reaches the right end of the image, it shifts down by one pixel and sequentially performs the processing from the left end in a similar manner, and repeats this until it processes to the lowermost right pixel and ends.

  FIG. 9 is a front view showing pixel values in which the enlarged view of FIG. 8 is represented by image data. A specific edge detection process according to the present embodiment will be described with reference to FIG. As is apparent from FIG. 9, there is almost no difference between the non-defect portion and the defect portion, and the change in the pixel data of the defect portion is moderate.

  FIG. 10 is a front view showing the result of processing the image data of FIG. 9 by the edge detection processing according to the prior art. A specific edge detection process according to the present embodiment will be described with reference to FIG. As is clear from FIG. 10, FIG. 10 shows the result of processing the pixel data of FIG. 9 by the conventional edge detection process.

  FIG. 11 is a front view showing a result of processing the image data of FIG. 9 by the edge detection processing according to the present embodiment. A specific edge detection process according to the present embodiment will be described with reference to FIG. In the example of FIG. 8, since the width of the defect is 12 pixels, the processing is performed using a pixel that is 1/4 pixel away from the processing target pixel as a peripheral pixel. Compared with FIG. 10 which is a result of the edge detection processing according to the conventional technique, a larger edge strength can be obtained.

Modification 1
FIG. 12 is an image showing image processing according to Modification 1 when pixel data for performing image processing is a characteristic value of pixel data within a predetermined vertical and horizontal rectangular range centering on each of the peripheral pixels. FIG. A specific edge detection process according to the present embodiment will be described with reference to FIG.

  In FIG. 12, G0 is a target pixel, and G1 to G8 are peripheral pixels that are separated from the target pixel by a predetermined number of pixels. The pixel values of the peripheral pixels G1 to G8 are not the values of the pixels themselves, but are characteristic values in all the pixels within a rectangular range having a predetermined size centered on each pixel. As this characteristic value, for example, an average, a maximum value, a minimum value, or the like of all pixels can be considered. In FIG. 12, for example, the average value of all the pixels within the broken line range is set as the peripheral pixel value at the location of the pixel G1. Similarly, the average values of the pixels G2 to G8 are obtained as peripheral pixel values at the respective pixel positions. The size of the rectangular range is determined in advance according to the roughness of the object to be inspected.

  In the above embodiment, the value of each pixel of the peripheral pixels is used as it is in performing the edge detection process, but this has the following drawbacks. When the object to be inspected is paper or the like, the surface is often rough. The roughness is uneven in the image data, and it cannot be distinguished from a defect having a small difference in shading in the difference in pixel value, and the roughness itself may be erroneously detected as a defect. According to the first modification, the unevenness of the image data due to the roughness is sparse and is much smaller than the defect. Therefore, in claim 2, the pixel values within the rectangular range are averaged to remove the roughness component and perform false detection. Can be reduced.

Modification 2
FIG. 13 is a front view of an image showing image processing according to the modified example 2 when the aspect ratio is different depending on the shape of the defect assumed in advance in each rectangular range centering on the peripheral pixels. A specific edge detection process of the present invention will be described with reference to FIG.

  In FIG. 13, G0 is a target pixel, and G1 to G8 are peripheral pixels that are separated from the target pixel by a predetermined number of pixels. The pixel values of the peripheral pixels G1 to G8 are not the values of the pixels themselves, but the average value of the pixel values of all the pixels within a rectangular range having a predetermined size centered on each pixel. When the defect assumed in advance is a vertically long line, the rectangle is long and vertically short as shown in the figure. If the line is long in the horizontal direction, the rectangular range is short in the vertical direction and long in the horizontal direction. The averaging of the rectangular range can remove the unevenness of the values on the image data due to the roughness, but at the same time, the edge component of the defective part is also removed to some extent, so that the accuracy of edge detection is lowered. This becomes prominent when it is a thin linear defect such as a scratch.

  In the process of the modification 2, the edge component that is removed is reduced by changing the aspect ratio according to the shape of the defect assumed in advance so that the ratio of the defective pixels in the rectangular range is increased. Thereby, the accuracy of edge detection can be improved.

  According to the defect inspection system including the image processing apparatus according to the present embodiment configured as described above, defects can be detected with high accuracy without increasing the size of the apparatus. In particular, when differential processing is performed on the image of the pixel of interest and the adjacent pixels, the difference between the defective portion and the non-defective portion is very small. Therefore, the edge strength obtained is also small. Therefore, by performing differentiation on the pixel of interest and a pixel separated from the position by an arbitrary distance, the edge strength obtained by increasing the difference between the two pixel values also increases. Thereby, the problem concerning a prior art can be solved.

  In the above embodiment and each modification, you may comprise only the image processing apparatus 3 in a defect inspection system or a defect inspection apparatus.

1 ... sheet products,
2 ... Camera,
3 ... Image processing device,
4 ... display device,
10 Image processing unit,
11: Image input unit,
12: Image storage unit,
13: Defect detection unit,
14: Address calculation unit,
15: Pixel data access unit,
16 ... Pixel data storage unit,
17. Pixel data calculation processing unit.

Japanese Patent Laid-Open No. 7-220058

Claims (5)

Image storage means for storing an image obtained by imaging the inspection object;
Image processing means for performing predetermined image processing on the image stored in the image storage means;
In an image processing apparatus provided with defect detection means for detecting defects based on the image subjected to the image processing,
The image processing means includes
Address calculating means for calculating an address for accessing pixel data stored in the image storage means for the image processing;
Pixel data access means for reading or writing pixel data of the image storage means based on the calculated address;
Pixel data storage means for temporarily storing the pixel data to be read or written;
Pixel data calculation processing means for rewriting the temporarily stored pixel data with the calculation result of the pixel data by the image processing;
An image processing apparatus that performs image processing on pixel data of a peripheral pixel separated by a predetermined number of pixels according to a target pixel that is an image processing target and a defect size assumed from the target pixel.   2. The image processing according to claim 1, wherein the pixel data for performing the image processing is a characteristic value of pixel data within a predetermined vertical and horizontal rectangular range centered on each of the peripheral pixels. apparatus.   3. The image processing apparatus according to claim 2, wherein each rectangular range centering on the peripheral pixels has a different aspect ratio depending on a defect shape assumed in advance.   A defect inspection apparatus comprising the image processing apparatus according to claim 1. Image storage means for storing an image obtained by imaging the inspection object;
Image processing means for performing predetermined image processing on the image stored in the image storage means;
An image processing method for an image processing apparatus comprising defect detection means for detecting a defect based on an image subjected to the image processing,
The image processing method executed by the image processing means is:
Calculating an address for accessing pixel data stored in the image storage means for the image processing;
Reading or writing pixel data of the image storage means based on the calculated address;
Temporarily storing the pixel data to be read or written;
Rewriting the temporarily stored pixel data with the calculation result of the pixel data by the image processing;
And a step of performing image processing on pixel data of peripheral pixels separated by a predetermined number of pixels according to a target pixel to be image-processed and a defect size assumed from the target pixel. Processing method.

Images