JPH01251288A - Picture processor for inspection system - Google Patents

Abstract

PURPOSE:To correctly execute the position and dimension matching of a product by correcting so that the phase and amplitude of the object picture data transformed by a transforming means and the reference data can be coincident. CONSTITUTION:Reference picture data and object picture data image-picked up by an ITV camera 21 are inputted through a picture input device 22 to a picture processor 23. Input picture data are Fourier-transformed by a frequency data transforming part 24 in the device 23 and inputted to a phase amplitude correcting part 25. At the correcting part 25, the correction coefficient is obtained by the Fourier transformed reference data, the correction coefficient is multiplied to the Fourier transformed object image data, and the amplitude and phase are made coincident to the reference picture data. The corrected object picture data are performed inverse Fourier transform by a frequency area data inverse transforming part 26, and the position and the dimension of the product are coincident to the reference picture. Further, since the difference calculation between the reference picture and the corrected object picture is executed by an inter-picture element arithmetic part 27 and when an abnormal part is present, it is detected, the position and dimension of the product can be correctly matched.

Description

[Detailed Description of the Invention] [Object of the Invention] (Industrial Application Field) The present invention is used for an inspection system that inspects products for the presence or absence of abnormalities such as scratches, and is an image processing system for inspection systems that performs inspection by image processing. to the processing equipment.

(Prior art) In general, inspection systems are installed on product production lines in the manufacturing industry depending on the product. However, when it is not possible to use chemicals, etc. for the product to be inspected, visual inspection or Inspection is performed using a system that combines an industrial television (ITV) camera, an X-ray device, an ultrasonic flaw detector, etc. with an image processing device. In this type of inspection system, when detecting an abnormality such as a scratch on a product or uneven printing on a printed matter, image data of a normal product without scratches or uneven printing is stored in advance, and the image data of the product to be inspected is After obtaining the image data, the difference between the pre-stored image data of the normal product and the image data of the product to be inspected is calculated, and it is determined whether the product to be inspected is normal based on the difference result.

FIG. 3 is a diagram showing an example of the configuration of a conventional image processing device for an inspection system. In FIG. 3, an image of a product taken by an ITV camera 1 is input to an image processing device 3 through an image input device 2, and is binarized by a binarization processing group 4 in an image processing device M3.

First, an image of an ideal product without any scratches or the like (hereinafter referred to as a reference image) is obtained using the ITV camera 1 and the image input device 2.

Next, an image of a product to be inspected (hereinafter referred to as a target image) is input using the ITV camera 1 and the image input device 2. Here, if the position of the product in the target image and the position of the product in the reference image are misaligned, the positions are corrected in alignment processing 15. The alignment processing unit 5 uses a number of techniques such as matching using fast Fourier transform (FFT), detecting differences in contour addresses, and detecting differences in centroids to find the positional deviation from the reference image, and performs alignment using affine transformation. I do. Also, if the dimensions of the product in the image differ when acquiring the target image data,
This is corrected in the dimension adjustment processing section 6. This correction is performed by detecting a difference between contour addresses and performing affine transformation on the target image based on the detection result.

After the image processing described above matches the position and dimensions of the product image in the reference image and the product image in the target image on the images, the difference detection section 7 calculates the difference between the two images,
The presence or absence of scratches on the product is detected from this difference result. This detection result is displayed on a display 9 through an image output device 8.

These processes are carried out by the central processing unit 10.
@ is being done. Further, a peripheral device 11, such as a data storage device or a printer, is used for processing other than the above-mentioned processing and for displaying the processing results, and this peripheral device 11 is also controlled by the central processing unit f10. In addition,
Each device is equipped with an image memory as required.

However, in conventional inspection systems such as those mentioned above,
For example, when a printed matter is to be inspected, it is difficult to clearly detect its outline because the 11-meter-degree image is delicate. That is, even if it is easy to align the reference image and the target image, it is difficult to align dimensions that differ by only a few pixels, which has been a major problem in automating inspection systems.

(Problems to be Solved by the Invention) As described above, in conventional inspection systems, if the outline of the product in the target image cannot be clearly detected, it is extremely difficult to match the position and dimensions with the reference image. Met.

SUMMARY OF THE INVENTION An object of the present invention was to solve the above-mentioned problems, and to provide an image processing device for an inspection system that can easily and reliably match the position and dimensions of a product in a reference image and a target image. The purpose is to

(Means for Solving the Problems) In order to solve the above problems, the image processing device for an inspection system according to the present invention converts the reference image data and the target image data into image data in a predetermined frequency domain, respectively. a converting means; a correcting means for correcting the phase and amplitude of the target image data converted by the converting means to match the phase and amplitude of the reference image data converted by the converting means; an inverse transformer that transforms the target image data inversely to the transformer, and calculates the difference between the target image data inversely transformed by the inverse transformer and the reference image data obtained by the imaging, and calculates the difference result. and detecting means for detecting the presence or absence of an abnormality in the target image data based on.

(Function) The inspection system image processing device configured as described above converts the reference image data and target image data into frequency domain image data, and calculates the amplitude and phase of the converted target image data based on the reference image data. By correcting the two images so that they match, the position and dimensions of the product can be aligned between both images, and scratches on the product can be detected by taking the difference between the reference image and the corrected target image. Can be done.

(Example of Solid Line) An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of an inspection system to which the present invention is applied. In FIG. 1, reference image data and target image data conveyed by an ITV camera 21 are input to an image processing device 23 through an image input device 22. In this image processing device 23, a frequency domain data converter 24 performs Fourier transform on the input image data, and inputs the Fourier transformed image data to a phase/amplitude corrector 25.

The phase/amplitude correction section 25 obtains correction coefficient data using the Fourier-transformed reference image data. Further, by multiplying the Fourier-transformed target image data by this correction coefficient data, correction target image data whose amplitude and phase match the reference image data can be obtained.

FIG. 2 shows the frequency domain data conversion section 2 of the image processing device W23.
4 and a block diagram showing the configurations of a phase and amplitude correction section 25. FIG. In FIG. 2, a Fourier transform circuit 41 performs Fourier transform on input image data. frame memory 4
2 stores reference image data. The ratio calculation circuit 44 calculates the ratio of the real part and the imaginary part of the amplitude of the Fourier-transformed reference image data, and the correction coefficient calculation circuit 45 calculates the ratio of the amplitude of 1''.
Correction coefficient data is obtained by internally dividing this ratio.

The frame memory 43 stores Fourier-transformed target image data, and the multiplication circuit 46 multiplies the target image data by the correction coefficient data obtained by the correction coefficient calculation circuit 45. Through such processing, the target image data is corrected.

The target image data corrected by the phase/amplitude corrector 25 is subjected to inverse Fourier transform in the frequency domain data inverse transformer 26, so that the position and dimensions of the product on the image match those of the reference image. In the pixel interoperation section 27, a difference calculation is performed between the reference image data stored in the frequency domain data conversion section 24 and the corrected target image data, and abnormal portions such as scratches are detected.

These processes are controlled by the central processing unit 30. Additionally, a peripheral device 31 that performs processing other than the processing described above and displaying the processing results, such as a data storage device,
Printers and the like are also controlled by the central processing unit 30. Note that each device is equipped with a memory for storing image data.

Next, the operation of this device will be explained.

First, image data obtained by conveying a normal product by the ITV camera 21 is inputted to the image processing device 23 through the image input unit 22 as reference image data. The reference image data input to the image processing device 23 is subjected to Fourier transformation in the Fourier transformation circuit 41 of the frequency domain data transformation section 24, and then temporarily stored in the frame memory 42. The ratio between the real part and the imaginary part of the amplitude of the reference image data stored in the frame memory 42 is calculated by the ratio calculation circuit 44, and 1" is internally divided by this ratio (m:n, m+n
-1) correction coefficient data is obtained by the correction coefficient calculation circuit 45.

Next, target image data is input in the same manner as the reference image data, subjected to Fourier transformation in the Fourier transformation circuit 41, and then stored in the frame memory 43. The target image data stored in the frame memory 43 is multiplied by a correction coefficient in a multiplication circuit 46, whereby the ratios of the real part and the imaginary part of the amplitudes of both image data are matched. The image data multiplied by the correction coefficient data in the multiplication circuit 46 is subjected to Fourier inverse transform in the frequency domain data inverse transform section 26, but at this time, the ratio of the real part and the imaginary part of the amplitude of both image data matches. Because there are
The amplitude and phase of the data to be inversely transformed match. This means that the position and dimensions of the product in the reference image and the target image have been aligned.

The difference between the reference image and the corrected target image is calculated by the pixel-to-pixel calculation unit 27, thereby detecting abnormalities such as scratches on the product, and the detected results are displayed on the display 29 through the image output device 28. .

Here, FFT is a typical method of Fourier transform, and the formula of this FFT is described as follows.

(k = N/2, . . . , (N/2)-1) where XK is a data string in the frequency domain, Xρ is a data string in the spatial domain, and N is the number of sample data.

When the phase changes in the above equation, it is transformed as shown in the following equation.

That is, by shifting the 2'th data in the frequency domain by the 2'th pixel, the phase change can be eliminated. As a result, this embodiment utilizes the fact that a deviation in the dimensions, position, etc. of a product on an image in the spatial domain becomes a phase deviation in the frequency domain. That is, after each of the reference image data and target image data is subjected to FFT processing, the ratio of the real part to the imaginary part of the amplitude of the reference image data is determined, and by shifting the target image data based on this ratio, the difference between both image data is Correct the amplitude and phase to match. Note that correction can be similarly performed using a Fourier transform method other than FFT.

Therefore, an inspection system using an image processing device can easily and accurately align and dimension the target image by correcting the phase and amplitude in the frequency domain of the target image based on the reference image. By taking the difference from the target image, it is possible to detect any abnormalities in the product.

The present invention is not limited to the above-mentioned actual example, and various modifications can be made within the scope of the gist of the present invention. for example,
In FIG. 1, the frequency domain data transformer 24 and the frequency domain data inverse transformer 26 may be the same. Moreover, the processing of each part can be performed by the same hardware by using, for example, a digital signal processing device and by changing the software. Furthermore, depending on the characteristics of the image of the product to be inspected, it is also possible to first perform binarization processing, rough positioning processing, etc., and then perform necessary processing.

[Effects of the Invention] As described above, according to the present invention, the positions and dimensions of products, etc. in the reference image and the target image are matched by correcting the phase and amplitude of image data in the frequency domain, and It is possible to provide image processing equipment for an inspection system that can easily and accurately match the position and dimensions of a product in a reference image and a target image even if the outline of the product is unclear.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an image processing device for an inspection system which is an embodiment of the present invention, FIG. 2 is a block diagram showing the configuration of the frequency domain data conversion section and the phase amplitude correction section in FIG. 1, Figure 3 is a drawing for a conventional inspection system.
FIG. 3 is a block diagram showing the configuration of a processed H image. 21... ITV camera, 22... Image input device, 2
4... Frequency domain data conversion unit, 25... Phase amplitude correction unit, 26... Frequency domain data inverse conversion unit, 27.
... Inter-pixel calculation unit, 28... Image output device, 29...
・Display, 30...Central processing VI device, 4
DESCRIPTION OF SYMBOLS 1...Fourier transform circuit, 42.43...Frame memory, 44...Ratio calculation circuit, 45...Correction coefficient calculation circuit, 46...Multiplication circuit. Figure 1 Figure 3

Claims (2)

[Claims] (1) An inspection that detects the presence or absence of an abnormality in the product to be inspected by comparing reference image data obtained by imaging a normal product and target image data obtained by imaging the product to be inspected. An image processing device for a system includes a conversion means for converting the reference image data and the target image data into image data in a predetermined frequency domain, respectively, and a phase and amplitude of the target image data converted by the conversion means by the conversion means. a correction means for correcting the converted reference image data to match the phase and amplitude; an inverse conversion means for converting the target image data corrected by the correction means in the opposite direction to that of the conversion means; Detecting means for calculating the difference between the target image data reverse-transformed by the converting means and the reference image data obtained by the imaging, and detecting the presence or absence of an abnormality in the target image data based on the difference result. An image processing device for an inspection system, characterized in that: (2) The correction means includes a calculation means for calculating the ratio of the real part and the imaginary part of the amplitude of the reference image data converted into frequency domain data by the conversion means; an acquisition means for acquiring correction coefficient data by internally dividing the ratio between the real part and the imaginary part of the amplitude by 1; and a target image in which the correction coefficient data obtained by the acquisition means is converted into frequency domain data. Image processing for an inspection system according to claim 1, further comprising means for multiplying the data to match the amplitude of the target image data with the amplitude of the reference image data converted into the frequency domain data. Device.