WO1997043623A1

WO1997043623A1 - Defect detecting apparatus and method

Info

Publication number: WO1997043623A1
Application number: PCT/JP1997/001562
Authority: WO
Inventors: Kimio Nakano; Satoshi Nishida; Yoshihiko Nakakoji; Toru Inomoto
Original assignee: Komatsu Ltd.
Priority date: 1996-05-10
Filing date: 1997-05-09
Publication date: 1997-11-20
Also published as: JPH09304292A

Abstract

An apparatus and a method for detecting a defect while discriminating uneven brightness which is not a defect from a real defect comprising a recess or projection, with a high accuracy. The brightness of each picture element of a picked-up image is binarized with a predetermined threshold value, and recessed and projecting defects, such as voids, and unevenness are extracted as portions where the brightnesses are lower than this predetermined threshold value. The amount of variation of the brightness of each picture element of the image is computed, the computed amount of variation of brightness of each picture element is binarized with a predetermined threshold value, and recessed/projecting defects, such as voids, and characters and circuit patterns printed on the surface are extracted as variations of brightness of a level higher than this predetermined threshold value. A recessed/projecting defect is detected by finding a region having a common coordinate position out of the regions extracted by these two kinds of threshold values.

Description

Description Defect detection apparatus and method

The present invention relates to a defect detection apparatus and method for capturing an image of a detection target and detecting an uneven defect appearing on the detection target. Background art

In the case of electronic parts produced by resin molding, such as IC molds, especially when the resin of the electronic parts solidifies, the resin surface becomes uneven due to uneven resin composition and damage to the mold. A small hole called) may occur.

These voids not only look bad, but if they reach the inside of the IC, they pose a major performance problem, so they must be reliably detected during the inspection process and treated as defective.

Also, in the case of a wafer chip or the like, irregularities may occur on a surface that should be originally smooth. If there is such an uneven portion, the wire may be bonded to the uneven portion in the wire bonding step, and poor bonding may occur. Therefore, it is necessary to reliably detect such irregularities appearing on the smooth surface in the inspection process and treat them as defective.

When an image of the surface of an electronic component having an uneven defect portion such as the above-described void is imaged, the defect portion is imaged as having a different brightness from the surroundings. In this case, in general, in the inspection process, the illumination of the inspection stage provided in the process, the position of the camera, and the like are adjusted, and the above-described irregularities such as voids have lower brightness than the surroundings. (Blackened).

Therefore, unevenness defects such as voids have been detected by binarizing the brightness of each pixel of the captured image adjusted in this way with a predetermined threshold value. In other words, if the low brightness area below the above-mentioned threshold value is a concave / convex defect such as a void (when the low brightness area exceeds a certain area), the above-mentioned predetermined threshold value is obtained. Brightness greater than Area 髙ぃ was determined to be normal.

However, the areas determined to be low in brightness below the predetermined threshold include non-defective surface irregularities in addition to voids, which are original defects.

Here, unevenness is a slight change in unevenness that appears on the surface of an electronic component such as an IC mold and is not a defect. If the normal surface is a mirror surface, the shallow, fine and subtle irregularities (none) will be uneven. On the other hand, if the normal surface is shallow and fine and delicate irregularities (plain ground), the mirror surface will be uneven.

Also, parts that are slightly inclined with respect to the normal electronic component surface are regarded as uneven. In addition, the dirt on the surface of normal electronic components is regarded as uneven. This unevenness may appear due to process differences between the manufacturing process equipment and the like.

Such uneven parts are imaged on a normal electronic component surface as having a different brightness than the surroundings. Therefore, as described above, when the illumination of the inspection stage and the camera position are adjusted so that the unevenness defect portion such as a void becomes lighter (blacker) than the surroundings, the unevenness portion also becomes the unevenness defect portion. Similarly, since the brightness is low, the unevenness may be erroneously detected as a “defect” and may be erroneously recognized as a defective product although it is a good product. Disclosure of the invention

The present invention has been made in view of such circumstances, and has as its object to clearly identify non-defective irregularities and original irregular-shaped defects so that defect detection can be performed accurately. Is what you do.

Therefore, in the main invention of the present invention, a defect detection apparatus that captures an image of a detection target and detects an uneven defect appearing on the detection target includes:

First binarizing means for binarizing the brightness of each pixel of the captured image with a predetermined threshold,

Differentiating means for calculating the amount of change in brightness of each surface element of the captured image,

Second binarizing means for binarizing the change in brightness of each pixel calculated by the differentiating means with a predetermined threshold value,

The first binarizing means provides a predetermined threshold value, brightness equal to or less than a predetermined value, and (2) a binarizing means for obtaining a coordinate position of a pixel having a brightness change amount equal to or greater than a predetermined threshold value, and a detecting means for detecting that there is an irregular defect at this coordinate position;

It is equipped with.

According to such a configuration, the brightness of each pixel of the captured image is binarized by a predetermined threshold value, and assuming that the brightness is equal to or less than the predetermined threshold value, irregularities such as voids and the like are considered. Are extracted.

Further, the amount of change in the brightness of each pixel of the captured image is calculated, and the calculated amount of change in the brightness of each pixel is binarized by a predetermined threshold value. As a result, irregular defects such as voids and characters and circuit patterns printed on the surface are extracted.

Therefore, an uneven defect is detected by obtaining an area having a common coordinate position among the areas extracted by these two types of thresholds.

As described above, according to the present invention, in the process of extracting various regions, uneven defects that should be originally detected and non-defect unevenness are clearly distinguished, so that the accuracy of defect detection is dramatically improved. BRIEF DESCRIPTION OF THE FIGURES

1 (a) to 1 (d) are plan views of an inspection object according to an embodiment of a defect detection device and method according to the present invention, and are diagrams used to explain a defect inspection processing procedure. ο.

FIGS. 2A and 2B are diagrams used to explain a defect inspection method capable of performing arithmetic processing at high speed.

FIG. 3 is a flowchart showing a processing procedure performed in the embodiment.

FIG. 4 is a flowchart showing a processing procedure capable of reducing the area for calculating the brightness change amount.

—This is a charter.

FIG. 5 is a flowchart showing a processing procedure capable of reducing the area for performing the brightness binary processing. BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of a defect detection apparatus and method according to the present invention will be described with reference to the drawings.

FIG. 1A is a plan view showing an IC mold 1 which is a defect inspection object assumed in the embodiment, and the surface of the IC mold 1 has a void 3 as a defect and a void 3 as a defect. Surface unevenness (dirt) 4 and printed characters “123” appear.

• First Embodiment

Hereinafter, a processing procedure for detecting the coordinate position of the void 3 will be described with reference to a flowchart shown in FIG.

First, as shown in FIG. 1 (b), an inspection area 2 to be imaged by a camera (not shown) is set on the surface of the IC mold 1 (step 101).

Then, the inspection area 2 is picked up by the camera as a light and shade image, and the brightness of each pixel of the picked-up image is binarized by a threshold C1 for distinguishing the void 3 from a normal part. (A high brightness area is binarized to logic "1" and a low brightness area is binarized to logic "0") (step 102).

As a result, an area of a pixel having a brightness equal to or less than the threshold value C1 (referred to as a blob) is obtained.

That is, as shown in FIG. 1 (c), the barycentric coordinate positions Gl (X1, Yl) and G2 (X2, Y2) of the blobs Pl and P2 are obtained, and these barycentric coordinate positions Gl and G2 are The numbers n (= 2) of the extracted blobs are stored in the above-mentioned memory while being stored in a predetermined memory in association with the blobs Pl and P2. Here, the blobs to be extracted are blob P1 indicating the body 3 and blob P2 indicating the surface unevenness 4.The printed character 5 has a high brightness and has a brightness equal to or higher than the threshold value C1. Is not extracted as a blob (step 103).

Next, a differential calculation process is performed to determine the amount of change in brightness (brightness gradient) of each pixel of the grayscale image of the inspection area 2 imaged by the camera (step 104).

Then, the brightness change amount of each pixel of the differentiated picked-up image is binarized by the threshold value C2 for discriminating between the void 3 and the normal state (step 105). Then, a blob having a brightness change amount equal to or greater than the threshold value C2 is obtained.

That is, as shown in Fig. 1 (d), the coordinates of the centroid of each blob P3, P4, P5, P6 Positions G3 (X3, Y3), G4 (X4, Y4), G5 (X5, Y5), G6 (X6, Y6) are obtained, and these barycentric coordinate positions G3, G4, G5, G6 are determined by the respective blobs P3, P4, It is stored in a predetermined memory in association with P5 and P6. Here, the extracted blobs are blob P3 indicating void 3 and blobs P4, P5, and P6 each indicating the printed character “123”. Regarding the surface unevenness (dirt) 4, since the boundary edge is not clear and the brightness change amount is low, the brightness change amount is smaller than the threshold value C2 and is not extracted as a blob (step 106).

Next, the barycentric coordinate positions G1, G2 of the blobs Pl, P2 stored in step 103 above and the barycentric coordinate positions G3, G4, G5, b3 of the blobs P3, P4, P5, P6 stored in step 106 above. The processing of determining that the distance is the same as that of G6 as void 3 is performed.

That is, first, i indicating each blob P i (i = l, 2) stored in step 103 is initialized to 1 (step 107), and the barycenter coordinate position Gi (i = l, 2 ) And whether or not the distance between the centroid coordinate position G3, G4, G5, G6 of each of the blobs P3, P4, P5, P6 stored in the above step 106 is equal to or less than a predetermined threshold value C3. . This threshold value C3 is a threshold value for judging whether or not the positions of the centers of gravity of both blobs coincide (step 109).

If the distance between the positions of the centers of gravity of both blobs is greater than the threshold value C3, i is incremented by +1 (step 1 1 1), and the processing of step 109 is executed until i reaches n ( Step 108).

As a result, as is clear from FIGS. 1 (c) and 1 (d), the distance between the barycenter coordinate position G1 of blob P1 and the barycenter coordinate position G3 of blob P3 is equal to or smaller than the threshold value C3. In the determination 109 (YES), it can be determined that the coordinate position G1 or G3 has the center of gravity of the void 3 (step 110).

In the case where there are two or more voids 3, as shown by a broken line D, a loop may be provided to shift from step 110 to step 111.

• Second embodiment

Next, an embodiment that can increase the processing speed by reducing the area for performing the differential operation processing will be described with reference to FIG. As shown in FIG. 4, in steps 201, 202, and 203, the brightness is binarized for the entire inspection area A, as in steps 101, 102, and 103 in FIG. 3, and the brightness becomes equal to or less than the threshold value C1. Blob Pl,? The process of storing the barycentric coordinate positions 01 and G2 of 2 is performed, but in the next stage, it is clear that either of the above blobs Pl or P2 is void 3, so the centroids of the blobs Pl and P2 are Differential calculation processing is performed only for the local area 6 of a fixed size B (B <A) around the coordinate positions Gl and G2, and the search for void 3 is performed by differentiating only the smallest area. ing.

That is, i indicating each blob P i (i = l, 2) stored in the above step 203 is initialized to 1 (step 204), and the processing of the following steps 206 to 209 is performed until i reaches n. Is repeatedly executed (steps 211 and 205).

That is, as shown in FIG. 2A, a rectangular local area 6 of a predetermined size B centered on the barycenter coordinate position Gi (i = 1, 2) of the blob P i is generated (step 206). ). Then, a differential calculation process is performed to determine the change in brightness (brightness gradient) of each surface element of the grayscale image of the local area 6 (step 207).

Then, the brightness change amount of each pixel of the local area 6 subjected to the differential processing is binarized by the threshold value C2 for discriminating the void 3 from a normal part (step 208). Then, a blob having a brightness change amount equal to or greater than the threshold value C2 is obtained. In this case, as shown in Fig. 2 (b), only the blob P3 indicating the void 3 is obtained, and the barycentric coordinate position G3 (X3, Y3) is obtained.The barycentric coordinate position G3 is obtained by the blob P3 And is stored in a predetermined memory. Note that blobs P4, P5, and P6 indicating the printed character "123" are outside the local area 6, and are not extracted as blobs. In addition, for the surface unevenness (dirt) 4, since the boundary edge is not clear and the brightness change amount is low, the brightness change amount is smaller than the threshold value C2 and is not extracted as a blob (step 209). .

Next, the one in which the distance between the center of gravity coordinate position G1, G2 of each blob Pl, P2 stored in the above step 203 and the center of gravity coordinate position G3 of each blob P3, stored in the above step 209, matches. Then, a process for determining that it is the void 3 is executed.

As a result, as is clear from Figs. 2 (a) and 2 (b), the barycentric coordinates of blob P1 Since the distance between the position G1 and the barycentric coordinate position G3 of the blob P3 matches, it can be determined that the coordinate position G1 or G3 is the barycenter of the void 3. Note that, as described above, without determining whether the distances match or not, when the determination YES in step 209 is reached (probe P3 detection time), it is determined that it is void 3. Yeah (Step 2 10).

If there are two or more voids 3, a loop that shifts from step 210 to step 211 may be provided as shown by a broken line E.

As described above, according to the second embodiment, the size B of the region to be differentiated is smaller than when the whole A is differentiated. In addition, since the search for the void 3 is performed only by differentiating the area around the low brightness area, the blob indicating the print character 5 is not extracted in the search process.

Therefore, arithmetic processing can be performed at high speed. In addition, a large-capacity memory is required to store an image obtained by differentiating the entire inspection area 2. However, according to this embodiment, the area to be stored can be reduced, so that a large-capacity memory is required. No memory is required. Therefore, the cost of the device can be reduced.

• Third embodiment

Next, contrary to the above-described second embodiment, the processing speed may be increased by reducing the area for performing the brightness binary processing.

That is, as shown in FIG. 5, in steps 301, 302, and 303, the process of binarizing the brightness change amount for the entire inspection area A by the threshold value C2 in the same manner as steps 101, 104, and 105 in FIG. Done.

Next, in step 304, a blob having a brightness change amount equal to or greater than the threshold value C2 is obtained.

That is, the blob P1 indicating the void 3 and the blobs P2, P3, and P4 respectively indicating the printed character "123" are extracted, and the center-of-gravity coordinate positions G1 to G4 of these blobs are stored. The number n (= 4) is stored (step 304).

In the next stage, since it is clear that any of the above blobs P1, P2, P3, and P4 is void 3, the centroid coordinate positions Gl, G2, G3, and G4 of blobs P1, P2, P3, and P4 Only a local area 6 of a fixed size B (B <A) around Search for void 3 by performing binarization using the threshold CI and binarizing only the smallest possible area.

That is, i indicating each blob P i (i == l, 2, 3, 4) stored in the above step 304 is initialized to 1 (step 305), and the following steps 307 are performed until i reaches n. -Repeat the processing of step 309 (steps 311, 306).

That is, a rectangular local area 6 of a predetermined size B centered on the barycenter coordinate position Gi (i = 1, 2, 3, 4) of the blob P i is generated (step 307). A process of binarizing the brightness of each pixel of the grayscale image with the threshold value C1 is executed (step 308).

Then, a blob having brightness equal to or less than the threshold value C1 is obtained. In this case, only the blob P5 indicating the body 3 is extracted. Blobs showing surface unevenness 4 are outside the local area 6 and are not extracted as blobs (step 309).

Then, the distance between the barycentric coordinate position Gl, G2, G3, G4 of each blob Pl, P2, P3, P4 stored in step 304 above and the barycentric coordinate position G5 of blob P5 stored in step 309 above A process that determines that the match is found to be a void 3 is executed.

As a result, the distance between the barycentric coordinate position G1 of the blob P1 and the barycentric coordinate position G5 of the blob P5 matches, so that it is possible to determine that the coordinate position G1 or G5 is the barycenter of the void 3. It should be noted that, as described above, without determining whether or not the distances match, at the time when the determination YES of step 309 is reached (at the time of detecting the probe P5), it is determined that it is void 3. (Step 310).

If there are two or more voids 3, a loop may be provided to shift from step 310 to step 311 as shown by the broken line F.

As described above, according to the third embodiment, the size B of the area where the brightness is binarized is smaller than when the entire A is binarized. Moreover, since the brightness 3 is binarized only in the area around the area where the brightness variation is high, the search for the void 3 is performed, and in the search process, blobs showing surface unevenness (dirt) 4 are extracted. Not done. For this reason, arithmetic processing can be performed at high speed.

In the above-described embodiment, it is assumed that a defect called void 3 is generated on the surface of the IC mold, but the inspection object to be inspected and the type of the defect are arbitrary.

For example, the present invention can be applied to the detection of unevenness defects appearing on a wafer chip, unevenness defects appearing on a ceramic substrate, unevenness defects on a circuit pattern, and concave defects appearing on the side surface of a laminated substrate.

In this case, since the brightness and the amount of change in brightness of the defect portion vary depending on the type of defect to be detected, the thresholds C l and C 2 may be varied according to the type of defect. ,.

In addition, the present invention is not limited to the detection of defects in electronic components, but can be applied to any inspection object as long as it detects irregular defects.

The following cases can be considered as modes for implementing the above-described embodiment.

That is, the processing contents of the above-described first to third embodiments may be appropriately stored in a floppy disk, and distributed and distributed to a user or an operator who performs a defect inspection. Further, the storage medium to be distributed and distributed may be a storage medium other than a floppy disk, such as a hard disk, an IC card, or a CD-ROM. In addition to the modes for distributing portable storage media as described above, hardware that uses the software of the present embodiment and software of the present embodiment are developed for distribution. Alternatively, a development computer containing the software may be communicably connected via a public line or a network, and may be distributed via the network.

At the time of the above distribution and distribution, in order to install the software on hardware (defect detection device) that uses the software, an installer may be attached and distributed to distribute the software. Data may be distributed and distributed _c Industrial availability

The present invention is not limited to the detection of defects in electronic components, and detects irregular defects. If applicable, it can be applied to any inspection object.

Claims

The scope of the claims

1. A defect detection device that captures an image of an object to be detected and detects an uneven defect appearing on the object to be detected.

Differentiating means for calculating the amount of change in brightness of each pixel of the 摄 image image,

A coordinate position of a pixel which has a predetermined threshold and a brightness equal to or less than a value by the first binarization means and a brightness change amount equal to or more than the predetermined threshold and value by the second binarization means. Detecting means for detecting that there is an irregular defect at this coordinate position;

Defect detection device equipped with

2. The predetermined threshold value of the first binarization unit and the predetermined threshold value of the second binarization unit are made different depending on the type of the uneven defect to be detected. 2. The defect detection device according to claim 1, wherein:

3. A defect detection device that captures an image of a detection target and detects uneven defects appearing on the detection target,

A storage means for storing a coordinate position of a pixel having a brightness equal to or less than a predetermined threshold and value by the first binarization means;

A differentiating means for setting a local area of a predetermined size including the pixels stored by the storage means, and calculating a change i in brightness of each pixel in the local area;

Detecting means for determining a coordinate position of a pixel having a brightness change amount equal to or greater than a predetermined threshold value by the binarizing means, and detecting that there is an irregular defect at the coordinate position. Defect detection device.

4. Irregular defects appearing on the detected object by capturing the image of the detected object In the defect detection device that detects

Differentiating means for calculating the change in brightness of each pixel of the captured image,

First binarizing means for binarizing the change in brightness of each pixel calculated by the differentiating means with a predetermined threshold value,

Storage means for storing, by the first binarization means, a coordinate position of a pixel having a brightness change amount equal to or greater than a predetermined threshold value;

A second binarizing unit for setting a local region of a predetermined size including the pixels stored by the storage unit, and binarizing the brightness of each pixel in the local region with a predetermined threshold value; Detecting means for determining a coordinate position of a pixel having a brightness equal to or less than a predetermined threshold value by the binarizing means, and detecting that the coordinate position has an uneven defect;

Defect detection device equipped with

5. A defect detection method for capturing an image of an object to be detected and detecting an uneven defect appearing on the object to be detected,

A first binarization step of binarizing the brightness of each pixel of the captured image with a predetermined threshold value;

A differentiating step of calculating an amount of change in brightness of each pixel of the captured image,

Change in brightness of each pixel calculated by the differentiating step! : A second binarization step of binarizing with a predetermined threshold value;

By the first binarization step, a coordinate position of a pixel having a brightness equal to or less than a predetermined threshold value and a brightness change amount equal to or greater than the predetermined threshold value is determined by the second binarization step, A detection step for detecting the presence of an irregular defect at the coordinate position;

Defect detection method comprising:

6. The predetermined threshold value of the first binarization step and the predetermined threshold value of the second binarization step are made different depending on the type of uneven defect to be detected. 2. The defect detection method according to claim 1, wherein the defect is detected.

7. A defect detection method for capturing an image of an object to be detected and detecting an uneven defect appearing on the object to be detected,

A first binarization step for binarizing the brightness of each pixel of the captured image with a predetermined threshold And

A storage step of storing a coordinate position of a pixel having a brightness equal to or less than a predetermined threshold value in the first binarization step;

A differentiating step of setting a local area of a predetermined size including the pixels stored in the storing step and calculating an amount of change in the brightness of each pixel in the local area; A second binarization step of binarizing a change in brightness of a pixel with a predetermined threshold value;

A detection step of determining a coordinate position of a pixel having a brightness change amount equal to or greater than a predetermined threshold value by the binarization step of 2 and detecting that there is an uneven defect at the coordinate position;

Defect detection method comprising:

8. A defect detection method for capturing an image of an object to be detected and detecting an uneven defect appearing on the object to be detected,

A first binarization step of binarizing a change in brightness of each pixel calculated by the differentiation step with a predetermined threshold value;

A storage step of storing a coordinate position of a pixel having a brightness change amount equal to or greater than a predetermined threshold value in the first binarization step;

A second binarizing step of setting a local region of a predetermined size including the surface element stored in the storing step, and binarizing the brightness of each pixel of the local region with a predetermined threshold value; ,

A defect detection step comprising: determining a coordinate position of a pixel having a brightness equal to or smaller than a predetermined threshold and a value in the binarization step of 2; and detecting that there is an irregular defect at the coordinate position. Method.