JP3002473B2 - Pattern inspection method and apparatus - Google Patents

Description

The present invention relates to a pattern inspection method and an apparatus therefor.

[Conventional technology]

2. Description of the Related Art There is a circuit pattern inspection apparatus, for example, an inspection apparatus for a printed circuit board, which detects a defect of a printed circuit board by imaging a printed circuit board, binarizing the printed circuit board, and comparing the binary image with reference data. The modulation of the input image signal obtained when an image of a defective portion such as a short circuit or disconnection of a pattern is taken is lower than that of a normal pattern portion due to deterioration of the high frequency resolution of an imaging system such as an imaging lens and an image sensor. Some conventional devices binarize an image signal of a defective portion whose modulation is reduced by moving a threshold value based on an input image signal. Related to this binarization circuit are computer graphics and image processing (1973), page 334 (COMPUTER GRAPHICS AND
IMAGE PROCESSING (1973) PP334).

FIG. 11 is a configuration diagram of a binarization circuit used in a conventional device and its operation. This binarization circuit delays a signal obtained by subtracting an offset value 900 from an input image signal 100 when a printed circuit board is imaged by a delay circuit 90 for an arbitrary time, as shown in FIG. After passing through the subtractor 91 where A <1), a signal obtained by adding the offset value 900 is input as a threshold value 901 to the comparator 92 together with the input image signal 100 and compared. Here, as shown in FIG.
When the input image signal 100 is larger than the threshold value 901, "1" (white),
When smaller, a binary signal 902 of "0" (black) is output.

[Problems to be solved by the invention]

The above prior art generates a threshold based on an input image signal,
Thus, the input image signal is binarized.

FIG. 12 shows an example in which an image signal of a minute defective portion with extremely reduced modulation is binarized using the binarization circuit. When the modulation attempts to binarize the image signal of a minute defect, binarization can be performed by increasing the amplitude range of the threshold (decreasing the attenuation rate A). In this case, FIG. ), The input image signal 100 and the threshold 901 follow very closely, so that the threshold 90
There is a problem that chattering 903 occurs in the binarized signal 902 at the boundary portion of the normal pattern due to the influence of noise at the portion where 1 intersects with the input image signal 100 as shown in FIG. This causes a false occurrence of a pattern defect or a frequent occurrence of false reports due to a deterioration in the inspection dimension accuracy of the pattern. Therefore, it is necessary to limit the amplitude range of the threshold value (increase the attenuation rate A). The image signal of the defective portion cannot be binarized.

SUMMARY OF THE INVENTION An object of the present invention is to provide a pattern inspection method and a pattern inspection method capable of reliably binarizing a minute defect without being affected by noise and detecting the minute defect.

[Means for solving the problem]

The object is to set a threshold value binarization circuit for performing binarization by comparing an image signal with a threshold value, a width of a defect that cannot be binarized by the binarization circuit to L, and a signal level P of a target pixel to be binarized.
(I) and the signal level P (i) of two pixels at a distance L / 2 and −L / 2 from the target pixel on a straight line including the target pixel.
+ L / 2) and P (i-L / 2), two one-dimensional first-order differential amounts P (i) -P (i-L / 2) and P (i + L / 2) before and after the target pixel.
-P (i) and the one-dimensional second derivative 2 * P (i) of the target pixel
−P (i + L / 2) −P (i−L / 2) are provided with a plurality of one-dimensional differential operation circuits so that the directions of the straight lines are different, and a one-dimensional secondary output from the plurality of one-dimensional differential operation circuits is provided. Two binarization circuits of a differential binarization circuit that binarizes to “1” if the sum of the differential amounts is negative and “0” if positive, and one of the one-dimensional differential operation circuits If the absolute value of the one-dimensional second derivative exceeds the set value and at least one of the two one-dimensional first derivatives has the opposite sign, the latter binarized output is selected. The present invention is achieved by a pattern inspection method and an apparatus including a selection circuit that selectively outputs a binarized output.

That is, in the present invention, an image signal is obtained by imaging a pattern to be inspected, and the image signal is binarized by a first binarizing means to obtain a first binarized image signal, and the image signal is differentiated. A second binarized image signal is obtained by performing binarization using the value obtained by the above-described process, and a binarized image is binarized by the first binarization unit using differential information of the image signal of the pattern to be inspected. The signal and the binarized image signal binarized by the second binarizing means are selectively switched to obtain a third binarized image signal synthesized, and the third binarized image signal is used. And detecting a defect of the pattern to be inspected.

According to the present invention, there is provided an image pickup unit for detecting a pattern to be inspected as an image signal, a first binarizing unit for binarizing a normal image signal detected from the image pickup unit, and an image signal detected by the image pickup unit. Second binarizing means for binarizing a very small image signal, selecting means for selectively switching and outputting the outputs of these binarizing means, and a binary image signal obtained from the selecting means. And a defect detecting means for detecting a defect by processing.

[Action]

FIG. 2 shows a signal waveform 101 of a pattern image including a defect. As shown in FIG. 9A, the image signal is binarized by the threshold binarization circuit to “0” when the image signal level is lower than the threshold 201 and to “1” when the image signal level is higher than the threshold 201, and usually excluding a minute defect 811. Threshold binarization output 120 where pattern 810 is binarized
(Shown in FIG. 6B). Also, as shown in FIG. 4, when the sum of the one-dimensional secondary differential amounts calculated in a plurality of directions by the differential binarization circuit is negative, “1”;
In the positive case, a differential binarized output 130 binarized to "2" is obtained. Here, the one-dimensional second derivative 251 takes a large absolute value in the extreme value of the signal waveform as shown in FIG. 3 and takes a large absolute value when the width of the signal waveform is L or less. In the image signal satisfying the condition, that is, the image signal of the defective pattern portion, the absolute value of the one-dimensional secondary differential amount becomes large. Further, the one-dimensional second derivative 251 takes a negative value when the signal waveform has a mountain shape and takes a positive value when the signal waveform has a valley shape, and therefore, the added value of the one-dimensional second derivative calculated in a plurality of directions. Is binarized to "1" when negative and "0" when positive, whereby the image signal of the defective pattern portion is surely binarized. If the absolute value of the one-dimensional secondary differential amount in a plurality of directions exceeds the threshold value 401 and at least one of the two one-dimensional primary differential amounts has an opposite sign, the selection circuit outputs a differential binarized output 13.
In order to select 0 and to select the threshold binarization output 120 and obtain the binarization result 140 for the other, a gradual extreme value 812 as shown in FIG.
(The absolute value of the one-dimensional second derivative is small), sharp edge 81
There is no misjudgment at 3 (the one-dimensional first derivative is the same sign). Therefore, the binarization sensitivity of the defect pattern portion can be increased regardless of the direction of the defect pattern and without affecting the binarization of the normal pattern.

〔Example〕

FIG. 1 is a block diagram showing one embodiment of the present invention. The image of the print pattern on the printed circuit board 1 forms an image on the sensor 3 after passing through the condenser lens 2. The analog image signal 100 of the print pattern obtained (imaged) by the sensor (imaging means) 3 is converted into an N-bit digital image signal 101 by the A / D converter 4 and then input to the binarization circuit 5. Is done. In the binarization circuit 5, the signal of the normal pattern portion is
There are a fixed threshold binarization circuit 20 for binarizing and a differential binarization circuit 30 for binarizing only a signal of a minute defective portion.
By appropriately switching and outputting the binarized image signals output by 20, 30 by the selection circuit 40, the two binarized image signals are combined. It is possible to inspect whether or not the wiring pattern on the printed circuit board has a defect by comparing the composite binary image signal with, for example, a reference binary image signal by a processing unit (not shown). it can.

FIG. 5 is an overall configuration diagram of a binarizing circuit 5 using one-dimensional differentiation in four directions. Digitized input image signal 10
1 is restored to a two-dimensional image by the extraction circuit 10 and an image signal 150 extracted from the central pixel of the extraction circuit 10
A fixed threshold binarization circuit 20 for binarizing the signal of the normal pattern portion, image signals 150, 151-a, 151-b, extracted from the central pixel and peripheral pixels of the central pixel used for differentiation.
152-a, 152-b, 153-a, 153-b, 154-a, and 154-b are input to a differential binarization circuit 30 that binarizes only a signal of a minute defect with high sensitivity. . The fixed threshold binarization circuit 20
The input image signal 101 and the fixed threshold 201 are compared by a comparator, and a fixed threshold binarized image signal 120 is output. Differential binarization circuit 30
Is the image signal extracted from the central pixel and peripheral pixels
150,151-a, 151-b, 152-a, 152-b, 153-a, 153-b, 154
−a, 154−b, sign of one-dimensional first derivative of input image 5
51-a, 551-b, 552-a, 552-b, 553-a, 553-b, 554-a, 55
A differential binarized image signal 130 obtained by binarizing the sum of the 4-b and the one-dimensional secondary differential amounts 251, 252, 253, 254 and the secondary differential amounts in four directions with a sign is output. The selection circuit 40 receives the input fixed threshold binary image signal 120 and the differentiated binary image signal 130
551-a, 551-b, 552-
a, 552-b, 553-a, 553-b, 554-a, 554-b and one-dimensional secondary differential amounts 251, 252, 253, 254 and the determination threshold 401, and one of the binary image signals 120 and 130 is used. And outputs it as a binary image signal 140 of the input image signal 101.

FIG. 6 is a configuration diagram of the cutout circuit 10. An input image signal 101 is output from a FIFO memory array 11 for delaying one horizontal scan and an image extraction circuit 12 using a shift register to an extraction terminal 13.
Is extracted as a two-dimensional image. As shown in FIG. 7, the cutout terminal 13 has a one-dimensional primary differential and a one-dimensional secondary differential consisting of a central pixel and two pixels on the same straight line, which are two pixels away from the central pixel by a distance L / 2. Are provided at positions arranged in four directions (0 ° template, 90 ° template, 45 ° template, 135 ° template). Pixel signals 150, 151-a, 151-b, 152-a, 152-b, 153-
a, 153-b, 154-a, 154-b are output to the differential binarization circuit. A binarized image signal output from the binarization circuit;
The pixel signal 150 of the center pixel 50 is output to the fixed threshold binarization circuit in order to match the timings of 120 and 130.

FIG. 8 is an overall configuration diagram of the differential binarization circuit 30. The one-dimensional secondary differential amounts 251, 252, 253, 254 in each direction obtained by the differential operation circuits 31, 32, 33, 34 are added by adders 61, 62, 63 to obtain the sum of the one-dimensional secondary differential amounts in four directions. . This means that if the sum is negative, it is sufficient to binarize it to “1”, and if it is positive, it is sufficient to binarize it to “0”. At the time of “1”,
In order to take a value of "0" at the time of positive, this is differentiated binary image signal 130
Should be output as Differential operation circuits 31, 32, 33, 34
551 of the one-dimensional first-order derivative obtained in each direction
−a, 551−b, 552−a, 552−b, 553−a, 553−b, 554−a, 554
−b is output to the selection circuit (judgment circuit) 40 as it is.

FIG. 9 is a specific example of the configuration of the 0 ° direction differential operation circuit 31, which is a pixel signal 150, 151-a, of the input image signal 101 extracted from the pixels 50, 51-a, 51-b for 0 ° direction differentiation. 151-b
Are input to the subtractors 64 and 65 and are subtracted. As a result, the one-dimensional primary differential amounts 161-a and 161-b on both sides are obtained, and the 0 ° direction sign bits 551-a and 551-b are output. Further, by subtracting the one-dimensional first-order derivative 161-a from the one-dimensional first-order derivative 161-b by the subtractor 66, a second-order derivative in the 0 ° direction is obtained. Output as 251. In the above, the differential operation circuit in the 0 ° direction has been described, but the configurations of the differential operation circuits 32, 33, and 34 in other directions (90 °, 45 °, 135 °) are the same.

FIG. 10 is a specific configuration example of the selection circuit 40. Differentiation 2
The 0 ° direction second derivative 251 input from the digitizing circuit 30 is
It is divided into a sign bit 351 and a numerical bit 451, and both are inputted to an exclusive-OR (EXCLUSIVE-OR) gate 71. To perform an exclusive OR operation between the sign bit 351 and the numerical value bit 451 means to find the absolute value of the second derivative expressed in two's complement format. The absolute value of the secondary differential amount thus obtained is input to the comparator 67 together with the determination threshold value 401. Here, the magnitudes of the two are compared, and when the absolute value of the second derivative is large, “1” is output as the absolute value signal 261. An exclusive OR (EXCLUSIVE-OR) gate 7
The output of 5 is the sign bit 551-a, 551
When -b has a different sign from each other (that is, when the image signal waveform is a mountain shape or a valley shape), it becomes "1", so that the 0 ° direction minute defect signal 271 has a large absolute value of the second derivative and the primary It becomes “1” when the original primary differential amount has a different sign. Less than,
The same applies to other directions (90 °, 45 °, 135 °).
The minute defect signals 271, 272, 273, 274 in each direction are input to the OR gate 83, and the multiplexer 41 is controlled by the replacement control signal 250. The multiplexer 41 outputs a replacement control signal.
When 250 is "1", the differential binarized image signal 130 is selected, and when it is "0", the fixed threshold binarized image signal 120 is selected. A differential binarized image signal is selected, and the two images are combined.

According to the above embodiment, the image signal with reduced modulation can be binarized without affecting the binarization of the normal pattern.

〔The invention's effect〕

According to the present invention, the optimum binarization condition can be given to both the normal pattern and the minute defect pattern. Therefore, the dimensional accuracy of the normal pattern of the binarized image and the binarization of the minute defect pattern are obtained. It is possible to provide a pattern inspection apparatus that can maintain both high sensitivity and high inspection reliability.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of a pattern inspection apparatus according to the present invention, FIG. 2 is a diagram showing a signal waveform of a print pattern image including a defect and a threshold binarized output thereof, and FIG. FIG. 4 is a diagram showing an example of binarization of an image signal, FIG. 5 is a diagram showing an entire configuration of the binarization circuit shown in FIG. 1, and FIG. The figure shows the entire configuration of the cutout circuit shown in FIG. 5, and FIG. 7 shows the structure shown in FIG.
FIG. 8 is a diagram showing the shape of the image signal differentiation template in the direction, FIG. 8 is a diagram showing the entire configuration of the differential binarization circuit shown in FIG. 5, and FIG. 9 is a diagram showing the 0 ° direction differentiation operation circuit shown in FIG. FIG. 10 is a diagram showing a specific configuration of the selection circuit shown in FIG. 5, FIG. 11 is a diagram showing the configuration and operation of a conventional binarization circuit, and FIG. FIG. 9 is a diagram for explaining an example of a disadvantage of the binarization method. DESCRIPTION OF SYMBOLS 1 ... Printed circuit board 2 ... Condensing lens 3 ... Sensor 4 ... A / D converter 5 ... Binarization circuit 10, 10 ... Cutout circuit 20 ... Fixed threshold binarization circuit 30 ... Differentiation binarization circuit, 40 selection circuit 31 0-direction differentiation calculation circuit 32 90-direction differentiation calculation circuit 33 45-direction differentiation calculation circuit 34 135-direction differentiation calculation circuit

──────────────────────────────────────────────────続 き Continuation of the front page (72) Inventor Norihiro Minatani 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Prefecture Inside Hitachi Video Engineering Co., Ltd. (56) References JP-A-63-64478 (JP, A) JP-A Sho 62-234466 (JP, A) JP-A-63-288565 (JP, A) JP-B 1-18623 (JP, B2) Jiko 62-20032 (JP, Y2) (58) Fields investigated (Int. Cl. ^7, DB name) G06T 1/00 H04N 1/04 103

Claims (5)

(57) [Claims] An imaging means for imaging a pattern to be inspected to obtain an image signal; a first binarizing means for binarizing an image signal of the pattern to be inspected obtained by the imaging means; Second binarizing means for obtaining a binarized image signal using a value obtained by differentiating the image signal of the pattern to be inspected obtained by Selecting means for selectively switching and outputting the signal from the first binarizing means and the signal from the second binarizing means; and selectively switching and outputting two signals obtained from the selecting means. A defect detection unit for processing the value image signal to detect a defect. 2. The pattern inspection apparatus according to claim 1, wherein said first binarization means binarizes the image signal by comparing a fixed threshold value with the image signal. 3. A pattern inspection apparatus according to claim 1, wherein said second binarizing means binarizes a sum of a plurality of second derivative values of said image signal by comparing the sum with a threshold value. . 4. An image signal is obtained by imaging a pattern to be inspected,
The image signal is binarized by a first binarizing unit to obtain a first binarized image signal, and the image signal is binarized using a value obtained by differentiating the image signal to obtain a second binarized image signal. An image signal is obtained, and a binarized image signal binarized by the first binarizing unit and binarized by the second binarizing unit using differential information of the image signal of the pattern to be inspected. And selectively switching the binarized image signal to obtain a third binarized image signal synthesized, and detecting a defect of the pattern to be inspected using the third binarized image signal. Pattern inspection method. 5. The pattern inspection method according to claim 4, wherein a defect of the pattern to be inspected is detected by comparing the synthesized third binary image signal with a reference value.