JPS6256974B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a flaw inspection device, and more particularly to a flaw inspection device capable of automatically determining the shape and type of flaws.

FIG. 1 is a diagram showing the overall configuration of a flaw inspection device. In FIG. 1, a laser beam oscillated from a laser tube 1 is focused on the surface of a sample 4 that is continuously moving in the X direction by the rotation of a rotating polygon mirror 2 and the action of a lens 3. Scan the entire area. Since the normal portion of the surface of the sample 4 is smooth, the reflected light of the laser beam is incident on the lens 3 instead. If there is a scratch on the surface of the sample 4, the reflection of the laser beam becomes scattered light, which excites the current in the phototube 7. The phototube 7 is a photoelectric conversion means that detects the presence or absence of diffusely reflected light and obtains a video signal. The signal processing device 8 detects the output signal of the phototube 7 and determines the presence or absence of a scratch.

If the sample surface is a metal-processed surface, isolated scratches are caused by foreign matter, voids, etc., and linear scratches are caused by cutter marks remaining during rough machining (lack of finishing machining) or during transportation. This is caused by rubbing etc. The former cannot be modified, but the latter can be modified by further finishing.

However, in conventional flaw inspection devices, flaws are detected by digitizing the output of the phototube 7, storing it in a huge memory, and processing it as a whole, so the type of flaw can be determined. It is impossible to process this in real time, and we have to rely on visual inspection by inspectors, making automatic judgment impossible.

SUMMARY OF THE INVENTION An object of the present invention is to eliminate the above-mentioned drawbacks of the prior art and to provide a flaw inspection device that can automatically determine the type of flaw in real time.

In order to achieve the above object, in the present invention, a two-dimensional pattern corresponding to the inspection surface captured by a phototube is sequentially cut out using small frames of an appropriate size, and a continuous flaw pattern within the small frames is cut out. The feature is that the presence or absence of scratches and their type can be automatically determined by checking the nature and size in real time.

The present invention will be explained in detail below. FIG. 2 is a diagram showing the relationship between linear scratches 9 and isolated scratches 10 and laser scanning lines i to l, and FIG. 3 is a diagram showing processing waveforms until the phototube output in FIG. 1 is digitized and stored. It is. In FIG. 3, the output a of the sensor 6 for detecting the scanning timing of the laser beam is generated before the laser scans the interval (S→E) between the pair of optical plates 5. After generation of sensor output a, Δt
After a second has elapsed, scanning gate b is opened, and the phototube output c during this period is binarized using a threshold value to obtain a binarized output gate d. Since N clocks e are generated while the scanning gate b is open, when this clock e is gated with the binarization power d, a binarization clock f is obtained.

Here, the binary clock is a binary pixel signal. In order to obtain this binarized clock, the video signal which is the output of the phototube is binarized by a binarization circuit using a predetermined threshold value, and then it is obtained by sampling. These are known.

FIG. 4 shows the two-dimensional pattern of the binarized clock corresponding to FIG. 2 obtained in this way.

In the present invention, an appropriate size (m
By scanning a frame of xn picture elements) and measuring the distribution of "1" within the frame at each scanning position, isolated flaws and linear flaws are determined. However, the frame moves from S to E in FIG. 4 and from i to l in a manner similar to commercial television scanning.

FIG. 5 is a diagram showing the configuration of a cutout circuit for scanning a two-dimensional pattern while cutting it out using this small frame. However, in FIG. 4, the size of the frame m×n is set to 4×4 for simplicity. This extraction circuit is composed of n-1 N-bit serial input type shift registers 50 and n m-bit serial input parallel output type shift registers 51, and operates with a binary clock f as input. The n shift registers 51 constitute the cutting frame 20.

Figures 6 to 11 are diagrams explaining the operation of this circuit.
In Figure 6, the clock is the fourth scanning line l (from the right)
The bit pattern of the extraction circuit at the time corresponding to "1" is shown. The clock advances sequentially and the scanning line l
Figure 7 shows the bit pattern at the time corresponding to the 12th "0". The bit pattern for the first "0" of the next scanning line is shown in Figure 8, the bit pattern for the second "0" is shown in Figure 9, and the bit pattern for the third "0" is shown in Figure 9. The bit pattern at the fourth "0" in FIG. 10 is shown in FIG. In this way, from the start point of each scan to the m-1th pixel, there are m×n (=4×4) picture elements and 2 pixels of the extraction circuit.
Since it does not match the dimensional pattern, the flaw determination described later is not performed during this time.

FIG. 12 shows a determination method for examining the presence or absence of scratches on the cutout frame 20 cut out in the above manner and the shape thereof. That is, it becomes as follows.

B. When there is no “1” around the frame (no contact).

The total value of “1” inside the frame is set as S, and the set value
Compare with S ₀ , if S is smaller than S ₀ , it is a good product.
If S is greater than S ₀ , it is considered an isolated flaw. Here, by appropriately setting the value of S ₀ , it is possible to allow minute scratches and to treat only large scratches as defective, or to treat even minute scratches as defective.

(b) When there are “1”s around the frame (there is contact).

Only when there is a “1” at the top and bottom of the frame (U〓D) and when there is a “1” at the left and right ends of the frame (L〓R) (opposite side contact), the sum of “1” in the frame Count the value S. The set value _S1 is compared with S, and if S is smaller than _S1 , it is judged to be a good product, and if S is larger than _S1 , it is judged to be a linear flaw. By setting S ₁ to an appropriate value, intermittent linear flaws can also be detected. Even if there are "1"s around the frame, if there is contact on the adjacent side, such as when there are "1"s at the left end and top end, or if only one end is in contact, it is unconditionally considered to be a good product. The reason for this will be explained with reference to FIG.

When the frame is at position a, it cannot be determined whether the group of "1"s is the size of the solid line portion or the dotted line portion shown at b, so no determination is made. (It is assumed to be a good product for the time being.) This determination is made when the frame comes to the position indicated by the dashed-dotted line b. The same thing applies when one side of c is in contact. At the position of the solid line with frame d, no determination is made because the adjacent side is in contact, but at the position of the dashed dotted line, the opposite side is in contact and a determination is made. In the case of an excessive scratch like e, the opposite side contacts U〓D, and if S≧S ₀ , it is determined to be a linear scratch.

FIG. 14 is a diagram showing an example of a determination circuit for carrying out the determination method of FIG. 13. In FIG. 14, each picture element around the cutting frame 20 is connected to the OR gates 11 to 14, and the signals U, D, R,
L becomes "1" when there is contact at each end.
Therefore, when the output of the AND gate 15 is "1", it indicates that there is a contact on the opposite side U〓D, and when the output of the AND gate 16 is "1", it indicates that there is a contact on the opposite side L〓R, and the OR gate 17 When the output is "1", it indicates that there is contact between opposite sides. Furthermore, when the output of the inverter 19 is "1" due to the operation of the OR gate 18 and the inverter 19, it indicates that there is no "1" around the cutting frame 20. On the other hand, the value S counted by the counting circuit 24 of "1" in the frame, the value S ₁ preset by the setting circuit 26, and the value S ₀ preset by the setting circuit 28 are compared to the comparison circuit 25 and 27, and when S is larger than the respective set values S ₁ and S ₀ , the comparison circuit output becomes "1". Therefore, the output of the AND gate 29 becomes "1" when there is a linear flaw, and the output of the AND gate 31 becomes "1" when there is an isolated flaw. These signals are further input to AND gates 41 and 42. This is provided so that the count circuit 41A becomes "0" from the start of scanning of the cutting frame 20 in each horizontal direction until the m-1th clock and does not perform any determination.

FIG. 15 is a diagram showing a circuit example of the counting circuit 24 of s in FIG.
becomes the input. Coefficient counter 21 inputs “1”
Outputs the number of bits 2 ⁰ , 2 ¹ , 2 ² . The two outputs of the coefficient counter are transferred to a 4-bit full adder 22.
Enter. The two outputs of the full adder 22 are further
It is input to the bit full adder 22. In this way, the sum S of "1"s in the m.times.n picture elements in the extraction circuit is latched in the latch 23.

FIG. 16 is a diagram showing an embodiment of the present invention. In this embodiment, the m-bit register in the extraction frame 3 is expanded to m+1 bits, and the "1" here is set as the m+1-th bit as the output terminal. By counting the number with a counter 32, the number B of "1"s lost during the movement of the cutting frame is counted. On the other hand, the number A of "1"s at the input end of the cutout frame 3 is counted by the counter 31, and this is the number of "1"s newly increased when the cutout frame 3 is moved. In this way, the number of connections from the cutting frame 3 may be 2n, and the counters 31 and 32 may also be extremely small.

The outputs A and B of the counters 31 and 32 are the subtracter 33
The difference A-B is output, added to the number S of "1"s immediately before the movement in an adder 34, and latched in a latch 35 as a new count value S.

Scan gate b is a signal that is turned on during the horizontal scan of FIG. 4, and clock e specifies each sampling time of that scan. Therefore, the output of the AND gate 36 is counted by the counter 37 and compared with the set value m of the setting circuit 39 by the comparison circuit 38. While the count value of the counter 37 is smaller than m, the right side of the cutting frame 3 protrudes from the two-dimensional pattern of FIG. AND gate 40 by the output of
Turn off and set B=0. Furthermore, since the sensor output a is output when the scanning of the machined surface by the laser beam turns around, that is, when one horizontal scan of the two-dimensional pattern ends and the next scan begins, this signal a causes the counter 37 and the latch to be output. 35
Clear the contents of and prepare for a new scan.

The counters 31 and 32 can be made relatively small even if configured as shown in FIG. 16, but if a ROM is used for these counter functions, the circuit can be made simpler. In other words, when the vertical direction of the cutting frame 3 is n bits, there are 2 ⁿ -1 or more addresses.
Prepare a ROM, for example, input end n of cutting frame 3.
Connect the bits according to the ROM address.
In other words, the n bits at the input terminal are regarded as binary, and the ROM address is specified using this. The contents of each address in the ROM include the address 2
The number of "1"s included in the decimal representation is stored. For example, an example where n=11 is shown in FIG. In this way, the number of "1"s is immediately output by specifying the address of the ROM.

As is clear from the above explanation, according to the present invention, relatively small cutting frames can be processed one after another in real time, so it is possible to automatically determine the presence or absence of scratches and their type, and also to automatically determine the presence or absence of scratches and the type of scratches. Since the dimensional pattern can also be stored in a register with a small storage capacity as explained in connection with the extraction circuit, there is an advantage that a large-capacity storage device is not required.

The wound shape determination method of the present invention can also be applied to automatic measurement of the concentration of red blood cells in blood, automatic measurement of the size of foreign substances mixed in a solution, etc.

[Brief explanation of the drawing]

Figure 1 shows the overall configuration of the flaw inspection device, Figure 2 shows the overall configuration of the flaw inspection device.
The figure shows how the sample surface is scanned, FIG. 3 is an explanatory diagram of the circuit operation for obtaining a two-dimensional digital signal pattern by scanning the sample surface, and FIG. 4 is a diagram showing an example of the obtained two-dimensional pattern. FIG. 5 is a block diagram of the extraction circuit, FIGS. 6 to 11 are explanatory diagrams of the operation of the extraction circuit, FIGS. 12 and 13 are diagrams showing a method for determining the presence or absence of scratches and their shape, and FIG. 13
A diagram showing an example of the configuration of a determination circuit that executes the method shown in the figure.
FIG. 15 is a diagram showing an example of a counting circuit for "1" within the cutout frame, FIG. 16 is a diagram of another embodiment of the counting circuit, and FIG. 17 is a diagram showing ROM contents. 1... Laser tube, 4... Sample, 7... Phototube,
8... Signal processing circuit, 9... Linear flaw, 10... Isolated flaw, 20... Cutting frame, 24... Counting circuit,
25, 27... Comparison circuit, 26, 28... Setting circuit, f... Binarization clock.

Claims (1)

[Claims] 1. A photoelectric conversion device that scans a sample surface with a laser beam, detects the presence or absence of diffusely reflected light from the sample surface, and converts it into a video signal; Binarize and sample to convert into a binary pixel signal 2
A digitizing circuit and a binary picture element signal obtained by the binarizing circuit are stored in an m×n memory means as a binarized image frame consisting of m×n picture elements by a group of shift registers. A cutting means for cutting out, a counting circuit for counting the damaged binary pixelized signal within the range of the binarized image frame cut out by the cutting means, and a binarized signal cut out by the cutting means. The first step is to calculate the logical sum of the binary pixelated signals with scratches cut out on the four surrounding sides of the image frame to detect the unblemished binary pixelated signals on the four surrounding sides. Comparing the count value S within the cutting frame counted by the gate means and the counting circuit with a predetermined set value _S0 , and when the value S exceeds the set value _S0 , it is determined that an isolated flaw exists. and a first determining means for determining the flawed binary pixelated signal cut out for each of the four surrounding sides of the binarized image frame cut out by the cutout means, separately for each of the four sides. Take the logical sum, take the logical product of the logical sum signals of the opposite sides of these logical sums, and then take the logical sum of these logical product signals, and create a damaged binary pixelated signal on the opposite side. second gating means for detecting;
In addition, when the comparison means compares the counted value S within the cutting frame counted by the counting circuit with a predetermined set value _S1 , and the value S exceeds the set value _S1 , a linear flaw exists. A flaw inspection device characterized in that it is equipped with a second determining means that determines whether