JPS619765A - Picture image processor - Google Patents

Abstract

PURPOSE:To eliminate the effects of picture image noises and to recognize a defect with high accuracy by comparing the difference of density between two adjacent area in the direction crossing the edge direction of plural picture elements with a prescribed discrimination level. CONSTITUTION:A differenitiating circuit 8 obtains the edge value corresponding to the difference of density between edges of a picture image and the extending directions of edges and stores them to a memory 9. At the same time, the edge linen is thinned to a line of a single picture element through a line thinning circuit 10 and stored to a frame memory 12 after the threshold value processing. Then the edge direction of the picture element is read out of the memory 9, and the difference of density is calculated between areas E1 and E2 crossing said edge direction. This calculated difference of density is compared with a prescribed discrimination level. If a large difference is obtained from this comparison, the flags are set up at a buffer repetitively until no picture element is detected. Thus it is decided that a matter to be checked has a defect if the flag has the larger value than the prescribed value.

Description

Detailed Description of the Invention Technical Field The present invention relates to an image processing system that captures an image of a detected object and processes the image data in order to detect defects such as irregularities, foreign matter contamination, and scratches occurring on the surface of the detected object. It relates to a processing device.

BACKGROUND ART There is a method of detecting shape defects such as scratches on the surface of an object to be detected by image processing. In the conventional image processing apparatus, an object to be detected is imaged by a television camera, and the image signal from the television camera is binarized. Then, the binarized image data is stored in an image memory, and the image data is subjected to arithmetic processing to detect defects in the object to be detected.

In the binarization process described above, since the amount of image data is small, it may be affected by electrical noise mixed in during the processing process or may not be able to obtain accurate data due to fluctuations in the amount of light received by the television camera. There was a problem that I couldn't do it.

Purpose The purpose of the present invention is to solve such conventional technical problems, to be able to obtain accurate image data without being affected by image noise or changes in the amount of received light, and to obtain accurate image data of the detected object. An object of the present invention is to provide an image processing device capable of detecting defects.

Embodiment FIG. 1 is a block diagram of an image processing apparatus 1 according to an embodiment of the present invention. The image processing device 1 irradiates light from a light source 2 onto the surface of an object to be detected 3 at an angle that does not cause regular reflection, and images the object to be detected 3 with a television camera 4 serving as an imaging means to capture the image as an image. An analog signal output from the television camera 4 is converted into a digital signal by an analog/digital converter 5. From the digitally converted image data, only the image of the detected object 3 is extracted by the binarization processing gate circuit 6.

FIG. 2 shows a binarized image capture screen P. The captured screen P is binarized by setting the image X portion of the detected object 3 to a logical value “1” and the other image portions to a logical value rOJ. The binarization processing circuit 6 operates so that only the image data of the image X part of the detected object 3 that is identified by the binarization is extracted, and the image data of the detected object 3 is stored in the first frame memory 7 and the differentiation circuit 8. is output. The image of the object to be detected 3 is stored in the first frame memory 7 as a grayscale image of, for example, 256 gradations, as shown in the third diagram.

The part indicated by the reference numeral ' is a defective part of the object 3 to be detected.

The differentiation circuit 8 extracts edges in the image by directional differentiation of the input grayscale image of the object to be detected 3.

To extract edges from a grayscale image, a differential method is generally used because it is sufficient to extract changes in density in the image, and this differential circuit 8 calculates the edge value corresponding to the grayscale difference in the edge of the image and the direction in which the edge extends. Desired. Data indicating the edge direction output from the differentiating circuit 8 is encoded and stored in the direction code memory 9. The edge line with continuous edges outputted from the differentiating circuit 8 is wide, so it is thinned to a line with a width of 1 pixel by the thinning processing circuit 10. The edge value is compared with a predetermined threshold value, and pixels having edge values below the threshold value are erased.

Through this processing, image data due to electrical noise mixed in during the image processing process is removed. The threshold-processed edge line drawing is stored as binarized data in the second frame memory 12, as shown in FIG.

In the second frame memory 12, the edge flag is 1-bit data indicating that if the logic value is "1", it is an edge point F1; if the logic value is "O", it is a non-edge point.

The data stored in the first frame memory 7, second frame memory 12 and direction code memory 9 is transferred to the data bus 1.
3, the processing device 14 performs processing for detecting defects in the detected object 3.

FIG. 5 is a flowchart showing processing operations for detecting defects in the object 3 to be detected. The processing device M14 performs so-called raster scanning from the upper left pixel to the lower right pixel so as to sequentially linearly scan the second frame memory 12 in the direction indicated by the arrow LS shown in the fourth diagram to detect edge flags. In step n1, it is determined whether or not the raster scan of the first frame memory has been completed. If the raster scan has not been completed, the process moves to step n2 and the edge flag of the threshold-processed image in the second frame memory 12 is detected. Next, the process moves to step n3, where it is determined whether an edge flag has been detected.

If no edge flag is detected, the process returns to step n1. When the edge flag is detected, the process moves to step n4, and the edge direction of the detected edge flag of pixel A is read out from the direction code memory 9. In step n5, the shading difference or shading ratio of the regions El and E2 intersecting the edge direction of the edge flag of the detected pixel A as shown in FIG. β is calculated. In terms of the number of pixels, the sizes of the regions El and E2 are, for example, about 3 pixels×3 pixels. The average density is calculated by reading out from the first frame memory 7 the density of the pixels corresponding to the addresses of the areas El and E2. and area El.

The difference or ratio of the average concentrations of E2 is determined. step n
6, it is determined whether the difference or ratio is greater than a predetermined discrimination level. When the difference or ratio is smaller than the discrimination level, the process returns to step n1. When the difference or ratio is greater than the discrimination level, proceed to step n7 and flag the buffer on the software.

The processing of the detected pixel A is now completed, and the edge flag of the pixel A on the second frame memory 12 is erased.

This is to avoid duplicate processing the next time the raster scan is performed. Next, a pixel connected to this pixel A is found, and if it exists, the difference in density or ratio of density in the adjacent area intersecting the edge direction is similarly determined for that pixel, and the difference or ratio in average density is determined as the discrimination level. , then the same buffer where pixel A was flagged is flagged. Then, the edge flag on the second memory 12 for that pixel is erased. In this way, similar edge flag detection processing is repeated one after another until no connected pixels are found or the average step difference or ratio between the two areas falls below the discrimination level. If the number of flags in the buffer set through such processing exceeds a predetermined value N, the object to be inspected 3 is determined to have a defect and is a defective product. In step n8, it is determined whether the number of flags in the buffer is N or more. If there are N or more flags, the process moves to step n9 and the inspected object 3 is determined to be defective. When the number of flags is N or less, the process moves to step nlo and the edge flag of the detected pixel is erased. In step nil, a pixel with an edge flag connected to the detected pixel is searched. In step n12, it is determined whether there are any pixels with edge flags. If a pixel with an edge flag exists, the process returns to step n4 and steps n4 to n12 are repeated. When there is no pixel with an edge flag, the process returns to step n1, and when the raster scan is completed, step n
At step 13, the object to be inspected 3 is determined to be a good product with no defects.

In this way, defects in the object to be inspected 3 are inspected, so that isolated points and image noise are not determined to be defective parts, and detection accuracy is improved.

As for the algorithms and circuits for the above-mentioned directional differentiation, line thinning processing, and threshold processing, it is sufficient to adopt appropriate methods that are well known in the past.

As another embodiment of the present invention, the level of density in two mask regions that are elongated in instant contact with each other in a direction crossing the edge direction of a pixel with an edge flag is level-discriminated, and the detected object 3 is
We will discuss the case of detecting defects in . 7) shows a state in which a grayscale image of the detected object 3 is stored in the first frame memory 7, and FIG. 8 shows a state in which the threshold-processed image of the detected object 3 is stored in the second frame memory 12. Indicates the stored state. The reference numeral indicates an image of a defective portion of the object to be detected 3, and the reference numeral FL indicates a pixel with an edge flag having a logical value of "1". The presence or absence of an edge flag is detected by raster scanning the second frame memory 12,
When an edge flag is detected, the edge direction of the pixel A is read out from the direction code memory 9, and a mask area M in one column as shown in Figure 91 is read out from the direction code memory 9.
is set. This mask area M has, for example, 9 pixels×
The size is one pixel. Therefore, the density of the pixel corresponding to the address of this mask area M is read out from the first frame memory 7, and the maximum and minimum values of the density within the mask area M are extracted. A flowchart showing the processing operation is shown in FIG. 1.0. Steps m1 to m4 correspond to steps n1 to n4 of the above-described embodiment, and similar processing is performed. In step m5, the maximum and minimum values of the density within the mask area M are determined as described above, and the difference or ratio between the maximum and minimum values is calculated. The process moves to step m6, where the difference or ratio is compared with a predetermined discrimination level, and if the value is smaller, the process returns to step m1. If the difference or ratio is greater than the discrimination level, the buffer is flagged in software. Pixel A detected with this
The edge flag of is cleared. When a pixel connected to the pixel A is detected, a line of mask areas is similarly set in the direction intersecting the edge direction, the maximum and minimum values of density are determined, and the maximum and minimum values are calculated. A difference or ratio is calculated. If this difference or ratio is greater than the discrimination level, the previously detected pixel A's flagged buffer is flagged and the edge flag for this pixel is cleared. Similar processing is repeated in this manner until no connected pixels are found or the pixels are interrupted, or until the difference or ratio between the maximum and minimum density values within the mask region M falls below the discrimination level.

If the number of flags in the buffer exceeds a predetermined value N in this process, it is determined that the detected object 3 has a defect and is defective. In step m7, a flag is set in the buffer, and the process moves to step m8, where it is determined whether the number of flags is N or more. If the number of flags is N or more, the process moves to step m9, where it is determined that the detected object 3 has a defect and is a defective product.

In step mlo, the flag of the detected pixel is erased, and in step m11, a pixel with an edge flag connected to the pixel with an edge flag is searched for.

If that pixel exists, the process returns to step m4 and steps m4 to m12 are repeated. If it does not exist, step m
Returning to step 1, when the raster scan is completed, step m13
The detected object 3 is determined to be a good product with no defects.

The embodiments described above are suitable for detecting defects in relatively large areas, and the other embodiments are suitable for detecting defects that are relatively elongated.

Effects Since the present invention is configured as described above, there is no need for positioning when automatically detecting defects in the object to be detected, which improves workability and eliminates light intensity fluctuations. The method achieves a certain level of detection accuracy and is virtually unaffected by image noise, and because processing is performed without reducing the amount of information in the image data, highly accurate defect identification can be achieved. , and has the effect that even defects of about one pixel can be detected as long as they are continuous.

[Brief explanation of drawings]

1 is a block diagram of an image processing device 1 according to an embodiment of the present invention, FIG. 2 is a diagram showing a binarized image, and FIG. 3 is a block diagram of an image processing device 1 according to an embodiment of the present invention.
FIG. 444 is a diagram showing the frame memory 7, and FIG. 6 is a diagram showing the second frame memory 12. ) is a diagram for explaining a region intersecting in the edge direction of a pixel with an edge flag, FIG. 53 is a flowchart for explaining the processing operation related to the present invention, and FIG. 7, FIG. 8 is a diagram showing the second frame memory 12 of another embodiment, and FIG. 9 is a diagram for explaining a mask area crossing the edge direction of a pixel with an edge flag in another embodiment. , FIG. 10 is a 70-chart for explaining the processing operation of another embodiment. 1... Image processing device, 3... Detected object, 4...
TV camera, 5...Analog/digital converter, 6
... Binarization processing gate circuit, 7... First frame memory, 8... Differentiation circuit, 9... Direction code memory,
DESCRIPTION OF SYMBOLS 10... Thinning processing circuit, 11... Threshold processing circuit, 12... Second frame memory, 14... Processing fU
]it, E 1, E2...area, M...mask area Agent Patent Attorney Kei Nishi Department Figure 6 (a) (b) Figure 9

Claims (2)

[Claims] (1) An imaging means for imaging the object to be detected, an analog/digital converter for converting an analog signal output from the imaging means into a digital signal, and a binarization process for the image data from the analog/digital converter. a binarization processing gate circuit that extracts only the image of the detected object within the imaging screen;
a first frame memory that stores a grayscale image of the detected object; and a differentiation circuit that directionally differentiates the image of the detected object and obtains an edge value corresponding to the grayscale difference of the image edge and a value indicating the direction in which the edge extends. , a direction code memory that stores the direction value obtained by the differentiation circuit as code data, a thinning processing circuit that thins the edge of the differentiated image, and a thinning processing circuit that thins the edge value of the thinned image. a threshold processing circuit that discriminates the level from the threshold value and erases pixels having edge values below the threshold; a second frame memory that stores the threshold-processed image; a processing device for processing the threshold-processed data, detecting edges of the threshold-processed image, retrieving edge directions of a plurality of pixels in which edges have been detected from the direction code memory, and detecting edge directions of the plurality of pixels in which the edges have been detected; Comparing the shading difference or shading ratio of two predetermined areas of a coherent shape adjacent in the intersecting direction with a predetermined discrimination level,
An image processing device characterized by comparing a set of regions having a discrimination level or higher with a predetermined size. (2) An imaging means for imaging the object to be detected, an analog/digital converter for converting an analog signal output from the imaging means into a digital signal, and a binarization process for the image data from the analog/digital converter. a binarization processing gate circuit that extracts only the image of the detected object within the imaging screen;
a first frame memory that stores a grayscale image of the detected object; and a differentiation circuit that directionally differentiates the image of the detected object and obtains an edge value corresponding to the grayscale difference of the image edge and a value indicating the direction in which the edge extends. , a direction code memory that stores the direction value obtained by the differentiation circuit as code data, a thinning processing circuit that thins the edge of the differentiated image, and a thinning processing circuit that thins the edge value of the thinned image. a threshold processing circuit that discriminates the level from the threshold value and erases pixels having edge values below the threshold; a second frame memory that stores the threshold-processed image; a processing device for processing the threshold-processed data, detecting edges of the threshold-processed image, retrieving edge directions of a plurality of pixels in which edges have been detected from the direction code memory, and detecting edge directions of the plurality of pixels in which the edges have been detected; The maximum value and minimum value of the density within the mask area extending thinly and adjacently in the intersecting direction are extracted, and the difference or ratio between the maximum value and the minimum value of the density is compared with a predetermined discrimination level, and the difference or ratio of the maximum value and the minimum value of the density is compared with a predetermined discrimination level. An image processing device that compares a set of regions with a predetermined size.