JPH01100676A - Video binarization method - Google Patents

Abstract

PURPOSE:To separate a background noise by successively updating a threshold level and taking the AND of a binary signal obtained by the new threshold level and the enlarging signal of the binary signal obtained by one time earlier. CONSTITUTION:An object 1 is image picked up by an image pickup device 2 to detect a binary picture based on an initial threshold TH0 preset to a threshold setting circuit 6 by a binarization circuit 5 through a multivalued memory 4. Then, the binary signal 9 through an AND circuit 8 writes a signal 14 enlarging the pattern of the signal 9 to a circumference to a binary memory 7 through the shift registers 11, 12 of an enlarging circuit 10. The AND of the output of the circuit 5 changing the threshold by DELTATH and the output of the memory 7 is taken in the AND 8 to repeat this operation and obtain the signal 9. Thereby, the good binary signal in which the background and the pattern are separated is obtained to the object which cannot obtain the good binary signal by a fixed threshold because of the ruggedness of a surface.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image binarization method in a famous device for inspecting patterns having uneven surfaces.

[Conventional technology]

When a video signal is binarized, noise may occur. Conventionally, as described in Japanese Patent Application Laid-Open No. 55-111131, there has been a method of repeatedly sampling video to improve the S/N,
As described in the 14th S, ICE Academic Lecture "Development of High-speed Automatic Visual Inspection Method for Circular Parts", there is a method of performing noise processing on the obtained binary image.

[Problem that the invention seeks to solve]

The above-mentioned conventional technology does not take into account video signals whose signal level changes significantly due to unevenness on the surface of the object, and does not provide good 2
It was difficult to quantify. That is, in the former method, since the signal itself is changing and is not noise, a good binary image cannot be obtained even if the video is repeatedly sampled. Furthermore, in the latter method, it is difficult to distinguish between defects and changes in signal level.

An object of the present invention is to provide a video binarization method that eliminates the drawbacks of the prior art described above.

[Means for solving problems]

The above object is to provide a binarization circuit that can sequentially raise or lower a threshold value, a means for enlarging or reducing a binary image,
Using a binary image obtained by enlarging or reducing the output of the above-mentioned binarization circuit, gating means is provided, which first detects a binary image using a preset threshold value, enlarges or reduces the binary image, and then This is achieved by repeating the process of detecting a new binary image by changing the threshold value of the binarizing circuit and gating the output of the binarizing circuit using the enlarged or reduced binary image.

[For production]

By successively changing the threshold of the binarization circuit, it is possible to detect the object as a binary image without being affected by changes in the signal level, but on the other hand, the background appears in the binary image,
It becomes noise. Therefore, in the present invention, the effective part of the binary image obtained by changing the threshold value is limited to the part determined by enlarging or reducing the reference binary image based on the binary image obtained by the preset threshold value. By repeatedly detecting a new reference binary image in such a way as to prevent the background from appearing as noise.

[Embodiments of the invention]

An embodiment of the present invention is shown in FIG. In FIG. 1, an object 1 is imaged by an imaging device 2, and the image signal 1 obtained from 2 is A/D converted by an A/D converter 3, and then
The data is input to a multilevel memory 4 having a capacity for a screen. Reference numeral 5 denotes a binarization circuit, which reads out the multi-level memory 4 and compares it with a threshold specified by the threshold setting circuit 6. The threshold value setting circuit 6 is given an initial value Tho and a threshold value change value ΔTH, and each time the output Thj to the binarization circuit 5 is binarized as a whole of the multilevel memory 4, ThL=Tho−ΔTHX (L
-1) (here i = 1.2...N) is calculated and updated. , 7 is a binary memory with a capacity for one screen, and the (at the time of image input), the entire memory is preset ('1' is written), and its contents are read out in synchronization with the operation of the binarization circuit 5. The outputs of 5 and 7 are ANDed by an AND gate 8, and a binary code 9 is output. Then, the binary code 9 is input to the enlarging circuit 10 and written into the binary memory 7 after being enlarged. When the processing for the entire multi-level memory 4, that is, for one screen is completed in this way, the threshold value setting circuit 6
In , the threshold value is updated, and the same process is repeated a set number of times N times, and the Nth binarization code 9 is used as the binarization code to be obtained.

Next, a specific example of the configuration of the enlargement circuit 10 will be explained. In FIG. 1, the binarization code 9 is passed through a shift register group 11 whose length corresponds to the length of the scanning line of the imaging device 2, and
71 serial-in-parallel-out shift register 12. By inputting the 12 multi-values to the OR gate 13, the enlarged signal 14 obtained by enlarging the pattern ('1' part) in the binary code 9 by n/2 pixels is obtained at the output of the 13. Can be done.

The above is an embodiment of the present invention. In FIG.
How the value code 9 changes will be explained using FIGS. 2 and 3.

In Fig. 2, the illumination light 15 is aimed at 4° as shown in (a).
.

When the object 1 is irradiated, diffuse reflection occurs due to the unevenness of the surface, and the video signal has a large level change as shown in (b), and it is not possible to obtain a good binary image by binarizing with a fixed threshold value. . FIG. 3 shows an example in which the image of such an object in FIG. 1 is binarized with N=3. As shown in Fig. 3(a), in the first binarization, the threshold Th□
= Tho, a part of the pattern to be detected (assuming that the pattern has a high reflectance and is detected brightly compared to the background) is output to the binarization code 9 as shown in (b). Then, the binarized code in (b) is expanded by the expansion circuit 10.

The binary code shown in (c) is written into the binary memory 7.

Next, in the threshold value setting circuit 6, Th2=Th□−Δ
By setting Th as a threshold value, the second binarization is performed, and the binarization signal as shown in (e) is output from the binarization circuit 5, and the content of the binary memory 7, that is, ( c)
is ANDed with the binary code 9, and the binary code 9 is output as shown in (f). And the binarization code of (f) is the expansion circuit 1
It is enlarged by 0 and the binary code shown in (g) is written into the binary memory 7. Next, in the threshold value setting circuit 6, T
The third binarization is performed by setting h3 = Th2 - ΔTH as the threshold, and the binarization as shown in (i) is performed.
By outputting the value signal from the binarization circuit 5 and ANDing it with the contents of the binary memory 7, that is, the binary signal of (g),
The desired binary signal 9 is obtained. As described above, with the threshold value Th technique in which the background does not appear as a binary signal (does not become '1'), a good binary image of the pattern cannot be obtained, and the entire pattern appears as a binary signal ('1'). In the case where the background appears as a binary signal (becomes 11') at threshold Th3 such as By performing an AND operation with the enlarged signal of the obtained binary signal, a good binary signal in which the pattern and the background are separated is obtained.

In addition, in the above explanation, an example was explained in which the enlargement circuit is used to spread the pattern to the surroundings and lower the threshold value.
It is clear that similar results can be obtained by changing the enlargement circuit of 8 to a reduction circuit, changing the gate circuit 8 from AND to OR, and increasing the threshold value.

〔Effect of the invention〕

According to the present invention, it is possible to obtain a good binary image by separating the background and pattern even for an object where the image signal changes greatly due to surface irregularities and a good binary image cannot be obtained with a fixed threshold value. Can be done.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an embodiment of the present invention, FIG. 2 is an explanatory diagram of a target example, and FIG. 3 is an explanatory diagram of a binarization process according to the present invention. 1... Target object, 2... Imaging device, 3... A/D converter. 4... Multi-value memory, 5... Binarization circuit, 6... Threshold value setting circuit, 7... Binary memory, 8... AND circuit, 9... Binary signal, 10... ...Expansion circuit.

Claims (1)

[Claims] 1. A means for binarizing the video signal from the imaging device, a means for changing the binarization threshold, and a means for extracting adjacent parts of the pattern from the binary image signal, and the method is determined using a preset threshold. A means is provided to create a binary image by enlarging or reducing the binary image signal with respect to the adjacent portion of the pattern, and to synthesize it with the binary image obtained by changing the threshold, and to convert the synthesized binary image signal into a new binary image. Binarize the video by creating a binary image by enlarging or reducing the adjacent parts of the pattern, and performing the process of combining it with the binary image obtained by changing the threshold value a predetermined number of times. A video binarization method featuring: