JPH0854220A - Appearance inspection method for object - Google Patents

Abstract

PURPOSE:To obtain an appearance inspection method for an object in which defect, e.g. a flaw or a foreign matter, on the surface can be detected easily and accurately and the presence of abnormality can be judged based on predetermined criteria. CONSTITUTION:In the appearance inspection of an object having a periodic pattern in the image recognition, color image of the object is taken in at least from the pattern region thereof (Step S1). The color image is then subjected to normalization of gray level (Step S2), monochromic processing through linear gray level conversion (Step S3), gray image processing (Step S4), masking (Step S5) and binarization (Step S6). Subsequently, it is subjected to inversion (Step S7), at need, and a pattern congruent with a reference pattern is removed from the processed image information (Step S8). Finally, presence of abnormality is judged based on the presence of a residual pattern (Steps S9-S11).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for inspecting the appearance of an object, and is particularly applied to the appearance inspection of an object such as a memory cell having a pattern having a periodicity in image recognition or a CCD (Charge Coupled Device) imaging cell. And a suitable inspection method.

[0002]

2. Description of the Related Art In an object having a pattern having periodicity in image recognition, for example, in a semiconductor device such as a memory cell or a CCD image pickup cell, there are about 200 memory cells per slice, depending on the type of IC. , 6 x 4.5 mm
It is about the size of a corner. When inspecting the presence or absence of a defect (abnormality) such as a scratch or a foreign substance in this memory cell, a human being visually looks for a scratch or a foreign substance on the memory cell with a microscope at a magnification of about 40 times. Therefore, it is necessary to inspect a considerable area at a magnification of about 40 times. This visual inspection is performed in the pellet check process of the previous process.

[0003]

As described above, since the conventional visual inspection of the memory cell is a sensory inspection by a microscopic operation by human eyes, a defect such as a scratch or a foreign substance on the memory cell is detected. The current situation is that the workers' pains are over the limit. In addition, there is a problem that a defect is likely to be overlooked due to human visual work, and there is a difference in judgment ability among individuals.

The present invention has been made in view of the above problems, and an object of the present invention is to accurately and easily find defects such as scratches and foreign substances on the appearance, and to detect abnormalities with a certain criterion. An object of the present invention is to provide a method for inspecting the appearance of an object, the presence of which can be determined.

[0005]

A method of inspecting an appearance of an object according to the present invention takes in an image of at least a pattern area of an object having a pattern having a periodicity in image recognition, and binarizes the taken-in image information. Processing is performed, patterns that are congruent with the reference pattern are removed from the patterns given by the image information after this binarization processing, and the presence or absence of an abnormality is determined by the presence or absence of patterns that remain after removal of this congruent pattern.

[0006]

In the above appearance inspection method, first, an image of at least the pattern area of the object to be inspected is captured. In order to convert the captured image information into data that can be used for determination, data having a large number of gradations is binarized with a certain threshold value. As a result, a portion brighter than the threshold value and a portion darker than the threshold value are separated. At this time, since the captured pattern has periodicity, by the binarization processing,
In the normal case, a set of patterns having a certain size and the same shape is formed, and in the case of an abnormality, there is a different-shaped pattern in addition to the same-shaped pattern. Therefore, the congruent pattern is removed based on a reference pattern of a predetermined size prepared in advance. As a result, all patterns disappear when normal, and irregular patterns remain when abnormal.
The part of the pattern where the residual image is formed becomes an abnormal part.

[0007]

Embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 2 is a block diagram showing the outline of an inspection apparatus for realizing the appearance inspection method according to the present invention. In FIG. 2, the object 1 to be inspected has a pattern having periodicity in image recognition. The object 1 is placed at a predetermined inspection position, and the entire image thereof is captured as a color image by an image capturing device 2 such as a CCD camera. The image capturing device 2 supplies the captured color image information to the image processing circuit 3. The image processing circuit 3 performs image processing such as density normalization, monochromeization by linear density conversion, gradation processing, binarization processing, and removal of certain patterns. After this image processing, the determination circuit 4 determines whether or not the pattern portion of the object 1 has a defect such as a scratch or a foreign substance.

Here, the normalization of the density is one of the methods carried out to reduce the influence of the images taken under different conditions, and the histogram of the image takes a certain form. The process of converting the density is performed.
Next, the linear density conversion will be described. When an image is captured by a camera, its density is generally distributed over only a small part of the total allowed density range. Now, assuming that a density histogram as shown in FIG. 3A is obtained, it is possible to enhance the contrast and make it easier to see by expanding the density distribution as shown by the arrow in FIG. 3B. This linear density conversion is one of the methods used when performing preprocessing.

Next, the processing procedure of the appearance inspection method executed by the inspection apparatus having the above-mentioned configuration will be described with reference to the flowchart of FIG. 1 by taking the case where an object to be inspected is a memory cell as an example. Here, the memory cell is a place for storing data in the IC memory 41 shown in FIG. 4, and has a configuration in which cells are arranged as a pattern having periodicity in image recognition. The size of the memory cell 42 is 6 mm × 4.5 mm at 1 Mbit.
It is a degree. The appearance inspection of the memory cell 42 is performed at a magnification of about 40 to 60 times, as in the case of the conventional visual appearance inspection.

In the flowchart of FIG. 1, for example, the entire image in the IC memory 41 shown in FIG. 4 is captured as a color image (step S1), and the color-normalized image is subjected to color normalization processing (step S2). By this normalization processing, image processing is performed so that a good-quality image for monochrome conversion can be obtained. Next, R (red), G
After linear density conversion is performed on each of the color signals of the three primary colors (green) and B (blue), monochrome processing is performed to obtain the sum of these (step S3), and then gradation processing is performed (step S4).

The normalization process, the monochrome process, and the shading process are processes for making defects such as scratches and foreign matters stand out. Therefore, when the appearance inspection of an object having a conspicuous defect is to be performed in advance, the above-described processing can be omitted. Further, it is not always necessary to capture the image of the inspection object as a color image, and in the case where the appearance inspection of the object that can easily distinguish the defective portion from other portions is performed, it may be captured as a monochrome image. . However, since scratches and the like may appear as colored defects, it is preferable to capture them as a color image because the defects can be surely recognized.

Next, a masking process is performed in order to cut out only the portion of the memory cell 42 which is the target portion of the visual inspection from the entire captured image (step S5).
As a result, the portion required for the visual inspection, that is, the memory cell 4
Only part 2 will be cut out. FIG. 5 shows the image after the masking process. In the drawings from FIG. 5 onward, it is assumed that the scattered area represents a black area and the castle area represents a castle area. Note that, when capturing an image, if the image of only the portion of the memory cell 42 is captured, the masking process can be omitted.

Next, in order to convert the data into data which can be judged, the data having a large number of gradations is binarized by a certain threshold value (step S6), and thereafter, inversion processing is carried out depending on the pattern of the cell portion. Perform (step S7). That is, an image that has been monochromeized and in which only the necessary portions have been extracted is composed of multiple gradations (for example, 256 gradations), and a value greater than this value with a certain value (for example, 130) as the threshold value is "1. , And the smaller one is set to “0”. As a result, a portion brighter than the threshold becomes "1" and a dark portion becomes "0". FIG. 6 shows the image after the binarization process, and FIG. 7 shows the image after the inversion process. In these figures, portions A and B indicated by arrows are defective portions.

By this binarization process, the image data becomes a set of patterns (figures) of a certain size and the same shape when the inspection target part is normal and has no defects such as scratches and foreign matters, and when there is a defect, it is abnormal. In addition to the patterns of the same shape, there are patterns of different shapes. Therefore, a process of removing a pattern congruent with the reference pattern based on a reference pattern of a predetermined size prepared in advance is performed (step S8). Specifically, a pattern (graphic) including a fixed number of dots is removed.

As a result, in the normal case, all the patterns disappear (the data are all "0"), and in the case of the abnormality, as shown in FIG. Includes "1"). That is, as is clear from FIG. 8, the pattern including the defective portions A and B is
Since it is not congruent with the reference pattern, only that pattern remains.

Therefore, it is judged whether or not there is a remaining pattern (figure) in the image after the removal of the congruent pattern (step S9). Process as if there is no defect (step S10), and if there is an abnormality,
Processing is performed assuming that the inspection target portion has a defect such as a scratch or a foreign substance (step S11).

In this way, it is possible to inspect whether or not the portion of the memory cell 42 has a defect such as a scratch or a foreign substance with a certain criterion. As a result, it is possible to solve a problem that a defect is overlooked or a judgment is different among individuals as in the case of a human visual work. If it is determined that there is an abnormality, the defect area is calculated from the coordinate position of the data of "1",
It is also possible to determine a position having an abnormality and automatically output the position data.

In the above embodiment, the case where the object to be inspected is a memory cell has been described as an example.
CC having a configuration in which cells are arranged as a pattern having periodicity
It may be a D imaging cell or the like. Also. INDUSTRIAL APPLICABILITY The present invention is not limited to the application to the appearance inspection of semiconductor devices such as memory cells and CCD image pickup cells, but can be applied to the appearance inspection of all objects having a pattern having periodicity in image recognition.

[0020]

As described above, according to the present invention,
In the visual inspection of an object having a pattern having a periodicity in image recognition, at least an image of the pattern area of the object is captured, binarization processing is performed on the captured image information, and the image after the binarization processing is performed. From the pattern given by the information, the pattern congruent with the reference pattern is removed, and the presence or absence of an abnormality is determined by the presence or absence of the pattern remaining after the removal of the congruent pattern.
It is possible to accurately and easily find a defect such as a scratch or a foreign substance on the external appearance, and further, it is possible to determine the presence or absence of an abnormality based on a certain criterion.

[Brief description of drawings]

FIG. 1 is a flowchart showing a processing procedure of a visual inspection method according to the present invention.

FIG. 2 is a configuration diagram showing an outline of an inspection apparatus for realizing an appearance inspection method according to the present invention.

FIG. 3 is a diagram for explaining the principle of linear density conversion.

FIG. 4 is an external view of a memory cell.

FIG. 5 is a diagram showing an image after a masking process.

FIG. 6 is a diagram showing an image after binarization processing.

FIG. 7 is a diagram showing an image after inversion processing.

FIG. 8 is a diagram showing an image after removing a certain size / pattern.

[Explanation of symbols]

 1 Object to be inspected 2 Image capture device 3 Image processing circuit 4 Judgment circuit 41 IC memory 42 Memory cell

Claims (3)

[Claims] 1. A method for inspecting the appearance of an object having a pattern having a periodicity in image recognition, comprising capturing an image of at least a pattern region of the object, performing binarization processing on the captured image information, It is characterized in that a pattern congruent with the reference pattern is removed from the pattern given by the image information after the binarization processing, and the presence or absence of an abnormality is determined by the presence or absence of the pattern remaining after the removal of the congruent pattern. Appearance inspection method for objects. 2. An image of at least a pattern area of the object is captured as a color image, and after normalizing the density, it is converted into a monochrome image and the density process is performed, and then the binarization process is performed. The appearance inspection method for an object according to claim 1. 3. The appearance inspection method for an object according to claim 1, wherein the abnormal position is also determined based on the coordinate position of the pattern remaining after the removal of the congruent pattern.