JPH0332734B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a defect detection method for optically scanning and detecting defects such as partial omissions and holes in a plate-shaped object.

When a plate-shaped object is passed between the light source array and the light-receiving element array, if the plate-shaped object has defects such as partial omissions or holes, how many times can the light from the light source pass without being blocked by the plate-shaped object? Therefore, it is possible to know whether there is a defect in the plate-shaped object from the output signal of the light receiving element. FIG. 1 shows the known principle configuration of this type of defect detection device.

In the apparatus shown in FIG. 1, n light sources 1a to 1
n light receiving elements 2a facing a light source array consisting of n
A light receiving element array consisting of ~2n is provided. In this example, n light sources and light receiving elements are provided in a 1:1 relationship, but the number of light sources may be different from the number of light receiving elements, and in extreme cases, the number may be one. The plate-shaped sample 3 moves in the direction of the arrow and passes between the light source array and the light receiving element array while blocking light. The output signals of the light receiving elements 2a to 2n are input to the signal processing device 4 and processed for defect detection. Sample 3
If there are defects such as missing parts 3a and holes 3b,
When these defective parts pass between the light source and the light-receiving element, light hits the light-receiving element, so that the output of the light-receiving element changes. Based on this change, the signal processing device 4
detects defects and outputs the detection results to the output device 5.

An example of a defect determination circuit provided in the signal processing device 4 is shown in FIG. This circuit includes a comparator 5, which compares the output signal S of the light receiving element 2 with a reference voltage So generated by a reference power source 6. If the characteristics of the light receiving element 2 are such that, for example, the output signal S increases as the light increases, the comparator 5 determines that there is a "defect" when S>So and outputs a signal of "1", for example. do.

However, such conventional defect detection methods have
There is a drawback that the detection accuracy depends only on the setting accuracy of the reference voltage So. In other words, during initial adjustment such as during equipment installation, the reference voltage So is adjusted for each light-receiving element to account for variations in the output characteristics of the light-receiving element due to manufacturing variations in the element or differences in the environment in which the element is placed. I have to do it. Further, when the output characteristics of the light receiving elements change due to aging, it becomes necessary to readjust the individual reference voltages So. These drawbacks may not be a big problem if the plate-like object being inspected is completely opaque, but cannot be ignored if it is semi-transparent or has a fine mesh structure (such as seaweed). .

An object of the present invention is to provide a highly accurate defect detection method that eliminates the above-mentioned drawbacks, is easy to adjust and manage, and is not affected by variations in characteristics among a plurality of light receiving elements.

In order to achieve this object, according to the present invention, defects in the sample are detected from the output data of each light receiving element obtained by optically scanning a plate-shaped object as a sample with a light receiving element array consisting of a plurality of m light receiving elements. In the defect detection method for detecting, all m pieces of data D(i) (where i=1, 2, ..., m) sequentially obtained from each light receiving element during one scanning cycle of the light receiving element array are stored. Then, the sum of m stored data _n 〓 ^i=
1 Calculate D(i), use the result to calculate the average value of m pieces of data and k corresponding to its constant k times, and calculate the average value D and the average value D for each stored data D(i). The absolute value of the difference |D(i)-D| is calculated, and the presence or absence of a defect in the sample is determined by comparing the magnitude of this |D(i)-D| with the k,
If there is even one case where |D(i)-|>k, it is determined that the sample is defective.

That is, in the following, the principle of the present invention will be explained in detail based on the graph of FIG.
The permissible range is expressed by k, which is a plus or minus threshold value based on the average value D. Now D(i)
Assuming (2), the decision value |D(i)-|, that is, |D(2)-|, corresponds to A in the figure. Therefore, if we compare the negative threshold value B of the hatched part, which is an allowable range, with the judgment value A, then A<B.
Therefore, the judgment value is within the permissible range and is not considered a defect. Next, if we assume that D(i) is the "defective part" D(4), the judgment value |D(4)-| corresponds to C in the figure, and is the positive threshold value B of the hatched part, which is within the allowable range. ’. The result is C>B', the determination value exceeds the threshold, and it is determined that there is a defect.

In this way, the present invention uses the average value of all data as the reference point for the threshold value, so it is not limited to data D(4), which is considered to be a defective part from the viewpoint of "large" amount of received light, but for example, D in the figure. As in (1), the amount of light received is too small, so it can be considered a defective part. This is effective when it is desired to detect deposits or the like on a translucent or semi-transparent object to be measured. In addition, as a result of the above effects,
The threshold constant k cannot be overlooked.

That is, the constant k determines the permissible range and is set in advance depending on the individual conditions in the measurement. These conditions include the accuracy of the equipment and the characteristics of the object to be measured. An example of the former is the variation in output due to the element being composed of a solar cell, and an example of the latter is the accuracy of the object to be measured. One example is the variation in output due to the use of "edible glue". In the example in Figure 4, k≒
By setting k to 0.5, variation in the output of each element was accommodated, but if this variation is small, it is preferable to decrease k to narrow the width of the tolerance range and increase the detection sensitivity of defective parts.

The above explains that the present invention is not affected by variations in output, but the present invention also takes into account aging of the element, and this is shown in Fig. 3b.
The explanation will be based on.

That is, this figure shows an example in which the output of the element shown in FIG. 3a has decreased overall due to aging. As above, taking D(4) as an example, the judgment value |D(4)−
It can be seen that | is also getting smaller. Here, if the threshold value is set as before, the output value D
The defect expressed in (4) will not be detected.
However, the threshold value in the present invention includes the average of the overall output values as a variable, and the output value of the element and the threshold value have a close relationship. Therefore, if the overall output of the element decreases, the threshold value also decreases accordingly, and the hatched area in the figure corresponds to the allowable range in this case. Therefore, as in FIG. 3a, D(4) is detected as a defective portion.

The present invention is constructed based on such a principle. Next, an example of an apparatus for carrying out the present invention will be explained based on FIG. 4. The light receiving element 11 inputs an output signal S corresponding to the input light to the A/D converter 12. On the other hand, a frequency divider 13 divides the frequency of the fixed-frequency reference pulse by 1/q to generate a system timing pulse. That is, frequency divider 13
outputs a start signal pulse T every q reference pulses to operate the system once every q reference pulses. By changing the value of q, the operating speed of the system can be adjusted, and in some cases, the frequency divider 13 can be omitted. The A/D converter 12 is started by a start signal T supplied from the frequency divider 13 via the AND gate 10, and when the conversion operation is completed, a conversion end signal is issued.
Outputs EOC and data signal D. These data signals D are input to m registers 21, 22, . . . and are sequentially stored. Control for sequential storage is provided by AND gate A1, which is provided in each register.
A2,... is performed.

The output pulses of the frequency divider 13 are counted by a counter 14 from an initial value of zero. The binary output signal X of the counter 14 is processed by the decoder 15 from 1 to m _+1.
is decoded into a control signal Y. Here, the i-th control signal turns on, that is, becomes a "1" signal only when the binary output of the counter 14 is equal to i. Therefore, the decoder 15 outputs "1" signals in order from 1 to m ₊₁ . Registers 21, 22,...
The i-th register has an AND gate A1,
Due to the action of A2, ..., the conversion end signal EOC=
Only when the i-th control signal=1, the value of the signal S that has been A/D converted by the A/D converter 12 is written. In other words, the data D(i) of the signal S that has been A/D converted the i-th time is stored in the i-th register, and finally the data of the signal S that has been A/D-converted m times is stored in the registers 21, 22, etc. will be written in chronological order. These m conversions correspond to the light receiving element 11 scanning from one end of the sample to the other end.

The next timing after completing the measurement of the signal S m times,
That is, the m ₊₁ output of the decoder 15 becomes "1", a series of operations related to A/D conversion are completed, and the system operation moves to the next stage. That is, adder 16 is started by control signal m ₊₁ , AND gate 10 is blocked, and counter 14 is held. Adder 16 is register 2
The sum of data D(i) written in 1, 22, . . . is calculated. Letting the conversion data of the A/D converter 12 be D(1), D(2), D(3), ...D(m) in chronological order,
The operation performed by the adder 16 is _n 〓 ⁱ⁼¹ D(i).

This output ΣD(i) of the adder 16 is multiplied by 1/m by a multiplier 17, where the average value of D(i) is determined, and the multiplier 18 multiplies it by a constant k/m. and here k/mΣD(i)=k{1/mΣ
D (i)}=k is calculated. Here, as described above, k is determined depending on individual conditions, and the calculation result is used as a threshold value for defect determination.

On the other hand, the difference D(i)- between each stored data D(i) of the registers 21, 22, . . . and the above average value is calculated by the subtracter 3.
1, 32, . . . and these differences are converted into absolute values |D(i)−| by absolute value forming circuits 41, 42, .

A comparison between the output signal k of the multiplier 18 and each output signal |D(i)−| of the absolute value forming circuits 41, 42, . . . is performed by the comparators 51, 52, . , if |D(i)-|>k, the output signal Bi becomes "1", and a "1" signal representing a defect in the sample is outputted as a system output via the OR gate 19. Note that the above calculations can also be executed by a microcomputer or the like.

The above circuit operation can be summarized as follows.

First, the sample is scanned from one end to the other, and the output of the light receiving element 11 is A/D converted m times. Let the conversion results be D(1), D(2),...D(m)=D(i). D(i) corresponds to the scanning result at a position of W·i/m, where W is the lateral distance (width) from one end of the sample to the other. After taking m pieces of data, the average value of the data =
1/mΣD(i) is calculated, and k, which is equivalent to multiplying it by a constant k, is calculated. On the other hand, |D(i)−| is calculated for each data D(i), and the result of the calculation is compared with k, and even if there is only one |D(i)−
If there is data D(i) such that D|>k, it is determined that the sample has a defect.

According to the present invention described above, by using the average value of the data D(i) and the constant multiplication k as the threshold value for defect determination, the influence of variations among multiple light receiving elements and changes over time can be completely eliminated. It is absorbed and has no effect on the detection results, so it has the advantage of not requiring complicated adjustment work for each light receiving element, although it is highly accurate.

[Brief explanation of drawings]

FIG. 1 is a layout diagram showing the basic configuration of a known defect detection device, FIG. 2 is a block diagram showing the basic configuration of a signal processing circuit, FIGS. 3a and 3b are graphs explaining the principle of the present invention, and FIG. FIG. 4 is a block diagram showing an example of an apparatus for implementing the present invention. 3...Sample, 11...Photodetector, 12...A/
D converter, 16... Adder, 17, 18... Multiplier, 21, 22... Register, 31, 32... Subtractor, 41, 42... Absolute value forming circuit, 51, 5
2... Comparator.

Claims (1)

[Scope of Claims] 1. In a defect detection method for detecting defects in a sample from output data of each light-receiving element obtained by optically scanning a plate-shaped object as a sample with a light-receiving element array consisting of a plurality of m light-receiving elements. , all m pieces of data D(i) (where i=1, 2, ..., m) sequentially obtained from each light receiving element during one scanning cycle of the light receiving element array are stored, and the stored m The sum of the data _n 〓 ⁱ⁼¹ D(i) is calculated, and the result is used to calculate the average value of the m data and k, which is multiplied by a constant k, and each of the stored data D( For each i), calculate the absolute value |D(i)-D| of the difference from the average value D, and compare the magnitude of this |D(i)-D| with the above k to determine the presence or absence of defects in the sample. A method for detecting a defect in a plate-like object, characterized by determining the defect.