JPS59138907A - Shape defect detector - Google Patents

Abstract

PURPOSE:To detect a fine shape defect easily and precisely by converting the output of an image pickup telecamera for a body to be inspected into binary picture signals successively on the basis of plural different slice levels, and storing it in a memory. CONSTITUTION:An image of the body 1 to be inspected is picked up by the ITV camera 2, whose output is converted into binary signals by N units of comparators 31-3N which have relatively close slice levels VS1-VSN. Outputs of the N units of comparators are stored as binary pictures in N units of binary picture memories 41-4N. Then, a defect detecting processor 6 consisting of a microcomputer, etc., operates switches S1-SN to access the binary pictures in the N units of binary picture memories 41-4N, deciding on a fine shape defect of the inspected body 1.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field] The present invention relates to a shape defect detector for detecting minute shape defects on the surface of an object.

[Background technology]

The conventional shape defect detector detects the object to be inspected as shown in FIG. Binarize and convert this binary signal into 2
The value image is stored in the image memory 4, and the shape measuring device 5V is connected to the 2
The quality of the inspected object J is determined by extracting characteristics such as width and length from the binary image in the value image memory 4 and comparing them with those of a non-defective item.

With such a shape defect detector, if the defects near the edges of the object to be inspected are minute, it is extremely difficult to set the slice level for binarization7z,! The average brightness changes due to changes in the color and reflectance of the object to be inspected (increase or decrease in illumination light, etc.), and correspondingly, it is impossible to binarize around the brightness of the defect. Ap, detection accuracy was low. Further, in order to match the slice level to the defective portion, leveling work by visual inspection JC was required, making it difficult to detect the defect.

[Purpose of the invention]

This invention makes it easy to remove minute shape defects! The object of the present invention is to provide a shape defect detector capable of detecting defects with high accuracy.

[Disclosure of the invention]

The shape defect detector of the present invention includes: a television camera that images a food object to be inspected; A plurality of binary image memories each storing the output sound of the numerical comparator as a binary image, and each of the binary images stored in the plurality of binary image memories is accessed to generate the object to be inspected. A defect detection processing device that performs defect detection processing for the ) 2 at i number of slice levels
By digitizing and obtaining multiple binarized images, multiple
Since any of the digitized images will always be binarized at a slice level near the brightness of the defect, defect detection can be performed easily. Furthermore, even if the defect in the object 1 to be inspected, the increase or decrease in illumination light, and the color and reflectance of the object 1 to be inspected change, an fc binarized image can be obtained by slicing around the brightness of the defect. Further, since there is no need to adjust the slice level according to the object to be inspected as in the conventional example, all microscopic defects can be easily detected.

Embodiments of the present invention will now be described with reference to FIGS. 2 to 4.

As shown in Fig. M2, this shape defect detector images the test food object J with an ITV camera 2, and sequentially outputs the output of the ITV camera 2 into a plurality of relatively close slice levels VS□ to VB□. The outputs of the N comparators 31-3N are stored as binary images in N binary image memories 41-4N, respectively.

-t, the defect detection processing device 6 consisting of a microconbeater etc. switches the switch S□~ to access each of the binary images in the N binary image memories 41 to 4N to detect the object to be inspected. ] All micro-shaped defects can be identified.

The operation of this shape defect detector will now be explained with reference to FIGS. 3 and 4 for VC in the case of N=4.

FIG. 3 shows a rectangular parallelepiped object to be inspected in this shape defect detector] Imaged by all ITV cameras 2, slice level vS1 ``82''S3 ''S4 (VS2 > VS
2VC3) V8. ) kW comparison! a31-34
The contour line L□ of the binary image obtained by binarizing with L2.
L3. L, 'e It was drawn on a flat plate. There is no need to extract the contour lines L-L4 for deletion.

In the above four binary images, the inside of the contour lines L□ to L4 has a white level (state of "1"), and the inside of the contour lines L□ to L4
The outside of the object is at a black level (1-OJ state). The contour line in Figure 3 ~L4 is getting larger sequentially because the edge of the object to be inspected has a radius, and this rounded part This is due to the fact that the brightness changes. Contour line I, 2. In L3, the bottom side is concave because there is a defect on the bottom side of the object to be inspected 1 and the brightness of that part is different from other parts. In addition, the ring 11s line L□,
In L4, the reason why there is no dent in the corresponding part of the bottom is due to non-JI at the slice level.

L f? :, 7) K, outline 3. It is necessary to detect the depression in L3. The defect detection processing device 6 detects this dent. This defect detection processing device 6 is designed to detect minute defects in the object to be inspected by comparing and discriminating two binary images that are binarized at the slice level of two pieces that are in instant contact with each other.

To explain concretely, for example, when comparing and determining binary images surrounded by contour lines L2 and L2, first, Figure 4 is used.
As shown in t, , r, the part of the black level (rOJ) around the boundary of the binary image surrounded by the inner contour line is changed to the white level (1'-IJ) and surrounded by the contour line L2. The enlarged binary image surrounded by the contour line L□′ is compared with the binary image surrounded by the contour aL2 shown in FIG. 4(B), and the difference is calculated. If extracted, the fourth
As shown in Figure (C), a binary image is obtained in which the normal portion disappears and only the defect 8ilP remains.

Defect detection processing device 6Vi, performs the same processing as above on contour a
L3. A binary image and a contour line L3. This is performed for the binary images each surrounded by L4.

As described above, in this embodiment, the object to be inspected 1 is imaged by the ITV camera 2, and the output of the ITV camera 2 is sequentially sliced and binarized at a plurality of relatively close levels. Binary painting 18! Because I wanted to get money,
By visual inspection of the entire slice level, minute shape defects can be automatically and accurately detected without adjustment.

〔Effect of the invention〕

According to the shape defect detector of the present invention, minute shape defects can be easily detected without the need for adjustment from slice level full visual observation Vc.

The disclosed technique will now be explained based on FIGS. 5 to 11.

FIG. 5 shows the entire configuration of a conventional sheet-like object inspection device. This sheet-like object inspection device uses illumination light from a light source]] which enters the thrift 12, and the light that has passed through the thrift 12 is transported through the entire imaging lens 13. imaged as. ITV camera 15
d. From the opposite side of the thrift 12, the line light H formed on the surface of the falling object 14 is imaged. Sheet-like object】4
If the surface of is flat, the line light H will be straight, but the fifth
If there is an uneven portion 14a as shown in the figure, the line light H is curved at that portion, and an image J in which a part of a straight line is curved is obtained as shown in FIG.

The video signal from ITV camera j5 is stored in frame memory 1.
6, one screen is recorded. Now, assuming that the image in FIG. 6 is an image recorded on the frame memory 6, the height A of the curve of this image J corresponds to the displacement of the uneven portion 14a. Paper cutting requires a lot of processing time, and it has been difficult to inspect the sheet 1-shaped object 14 online. It was abandoned. Since the frame memory 16 is used, the cost is extremely high.

FIGS. 7 and 8 show a perspective view and a block diagram of a sheet-like object inspection device that can eliminate the above-mentioned drawbacks, can be easily put online, and is inexpensive.

As shown in FIG. 7, this sheet-like object inspection device uses a diffuser plate (for example, a frosted glass) to direct illumination light from a light source 2 to a 22V
'The sheet-like object 23 with the CJ rainbow dispersed to some extent.
1j1.2] and installed on the opposite side of the
The entire surface of the sheet-like object 23 is imaged by the TV camera 24. That is, ITV camera 24Irr,...
The specular reflection image of the light source 2J will be captured, but the diffuser plate 2
2, the light source 2] has an elliptical luminous intensity distribution whose central phase is open and gradually darkens in the radial direction.

Then, an image having brightness corresponding to the luminous intensity distribution of the light source 21 as described above is obtained from the ITV camera 24. As shown in FIG. 8, the video signal corresponding to this image is sequentially divided into slice levels US
If the binarization circuits 25 to 29 perform binarization, the inside of the contour lines LL□ to LL5 as shown in FIG.
These five binarization circuits 25 will yield five binary ii!11 image numbers corresponding to a binary image (in a normal case) with black level C (J) on the outside (normal case). The standard pattern output from the pattern memory 30 which stores an inverted pattern of the standard pattern PP of the binary image (the diagonal is the black "0" level) as shown in FIG. Exclusive OR gate 31 inputted with binary image signal
~35, a total counter 36~ for the number of "1" level binary signals output from the exclusive OR gates 31~35.
40, respectively, to find the correlation between the binary image obtained from the five binarization circuits 25 to 29 and the standard pattern, that is, the number of pixels in the portion where both match, and determine the degree of the correlation, That is, the flatness, reflectance, etc. of the surface of the sheet-like object are all inspected by determining the count values of the counters 36 to 40 using the microcomputer 4J.

Slice levels U8□-US2 of binarization circuits 25-29
If the surface of the sheet-like object 23 is normal, the contour line LL~LL5 consisting of concentric similar ellipses as shown in FIG. A binary image with the white part surrounded is obtained, pattern memo') 3
0 stores the inverted pattern of the standard Hahn shown in FIG. Null that it becomes the largest and its value exceeds the set value. The microcomputer 41 detects that the count value of the counter 39 is the largest and exceeds the force setting value, and determines that the surface of the sheet-like object 23 is normal.

However, if the surface of the sheet-like object 23 is uneven and the contours HLL-LL-5 are distorted as shown in FIG. The count values of the counters 36 to 40 all exceed the set value, and the microcombi-1-2 detects this and determines that the surface of the sheet-like object 23 is uneven. do.

Moreover, the reflectance of the surface of the sheet-like object 23 is
As shown in Figure 1 (■), if the overall V of the contour line LL - LL5 becomes smaller, for example, the contour line LL2 indicates δi.
When the correlation between the L binary image and the standard pattern is the strongest, the count value of the counter 37 becomes the largest and exceeds the set value.

The microcomputer 41 controls the counter 3 in the normal case.
It is detected that the count value of the counter 37 different from 9 is the closest and exceeds the set value, that is, it is detected that the counter 37 whose count value is the closest and exceeds that value or the set value is not 39. Thus, it is determined that the reflectance of the surface of the sheet-like object 23 is different.

As a result of this configuration, the surface condition of the sheet-like object 23 can be easily and inexpensively inspected without requiring a frame memory. Furthermore, the processing of the microconveyor 41 is simple, the processing time is known, and online inspection (inspection while transporting the sheet-like object 23) is possible.

[Brief explanation of the drawing]

Fig. 1 is a block diagram of a conventional shape defect detector, Fig. 2 is a block diagram of an embodiment of the present invention, Fig. 3 is an explanatory diagram showing the outline of a binary image, and Fig. 4 is a diagram showing the outline of a binary image. , (Q is an explanatory diagram for explaining the defect detection processing method, M5 is a configuration diagram of a conventional example of the disclosed technology, FIG. 6 is an explanatory diagram showing the image obtained by this method, Figure 7 is a configuration diagram of an improved example, Figure M8 is also a Bunk diagram, Figure M9, Figures 10 and 11 (A), (The odor is an explanatory diagram of the inspection method.) Object to be inspected, 2.・ITV camera, 3J~3N...
Comparator, 4'' to 4N... Binary image memory, 6... Below defect detection processing device 1 Figure 2 Figure 73 (A) (B) /---------) 1 1 ] \- --7Mo□---ノ'P (C) / ) 14a 14 Fig. 5

Claims (1)

[Claims] A television camera that images the object to be inspected, multiple comparators that sequentially binarize the output of the television camera at different relatively close slice levels, and each of the outputs of this number of comparators is binarized. fj to be stored as an image! number 2
Value image memory and this plurality of binary values II! A shape defect detector comprising a defect detection processing device which accesses each of the binarized images stored in a II image memory and performs all defect detection processing for the above class inspection object.