JP2660560B2 - Defect inspection method - Google Patents

Description

Description: FIELD OF THE INVENTION The present invention relates to a method for inspecting a periodic pattern of an industrial product for defects.

[Conventional technology]

Industrial products with periodic patterns include shadow masks for CRTs for color televisions, color separation filters for color imaging devices, color filters for liquid crystal display panels, mesh electrodes for electron tubes, VDT filters, mesh filters for filtration devices, and rotary encoders. , Linear encoder, IC
Photomask, Fresnel lens, lenticular lens, etc. In some cases, unit patterns having certain optical properties and shapes are regularly and repeatedly arranged in a one-dimensional direction or a two-dimensional direction. The unit patterns are arranged while the arrangement pitch in the dimensional direction gradually changes. In such an industrial product, defects caused by non-uniformity of optical properties due to the shape, size, position, and the like of the periodic pattern become a problem. Therefore, as a method of inspecting this defect, there are proposed, for example, Japanese Patent Application Laid-Open Nos. 61-256237 and 62-201335.

The conventional periodic pattern inspection method generally measures elements such as the shape, size, arrangement pitch, or color of a unit pattern constituting the periodic pattern with a dedicated measuring instrument, and uses the naked eye or a microscope. Therefore, there are problems in inspection efficiency (production efficiency), inspection accuracy, reliability, and the like. On the other hand, in the above proposal, image data is divided into a plurality of pixels by an image input device to obtain image data in units of frames, and random noise, shading components, noise, etc. are removed therefrom to judge whether the product is good or bad. This makes it possible to automatically determine good or bad,
Improving reliability and production efficiency.

[Problems to be solved by the invention]

However, in the above-described method, for a captured image data, a random noise component generated at a stage of obtaining the image data, a shading component caused by a luminance distribution of a light source, and a pulse characteristic caused by dirt attached to an optical system or a product. After performing predetermined processing, such as removing fixed noise such as noise, a defect is detected by comparing it with a predetermined threshold.If there is uneven illumination or resolution distribution, a certain sensitivity within the screen There is a problem that it cannot be obtained. Further, when the work (sample to be inspected) has a short-period change in transmittance or the like, there is a problem that many such changes are detected as defects.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and has as its object to provide a defect detection method that stabilizes detection sensitivity and reduces detection unevenness.

[Means for solving the problem]

Therefore, the present invention relates to a defect inspection method for inspecting a periodic pattern of a sample to be inspected for defects in a periodic pattern. It is characterized in that a detection threshold distribution is obtained by applying a certain threshold value by performing a smoothing filter process on the minute signal and a defect is detected by comparing the detection threshold distribution with the variation signal.

[Action]

In the defect detection method of the present invention, a detection signal is obtained by extracting a variation signal caused by a defect or period unevenness from image data obtained by photographing an inspection area, performing a smoothing filter process on the variation signal, and adding a certain threshold. Since the distribution is obtained, the detection threshold distribution becomes large in a portion having the periodic unevenness, but becomes relatively small in a defect. Therefore, when the defect detection processing is performed using such a threshold distribution, the portion of the periodic unevenness is not detected as a defect, and the detection accuracy can be improved.

〔Example〕

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a diagram for explaining one embodiment of a defect detection method according to the present invention, and FIG. 2 is a block diagram showing one example of an apparatus for implementing the defect detection method according to the present invention.

Before describing the defect detection method of the present invention, an outline of the apparatus will be described first with reference to FIG.

In FIG. 2, 1 is a camera, 2 is an image processing unit, 3 is a control unit, 4 is a display, 5 is a driving mechanism, 6 is a sample to be inspected, 7 is a diffusion plate, 8 is a lamp, 9 is a power source, and 10 is Shows the XY stage. The power supply 9, the lamp 8, and the diffusion plate 7 constitute a so-called transmission illumination unit. The lamp 8 is turned on by the power supply 9, and the sample 6 to be inspected is illuminated through the diffusion plate 7 by the emitted light. The camera 1 captures an image of the inspection area of the sample 6 to be inspected in this illuminated state. The XY stage 10 moves the inspection area of the sample 6 to be inspected.
Control drives the XY stage 10. Image processing unit 2
Is for inputting a video signal captured by the camera 1, A / D converting the video signal, storing the video signal in a frame memory in association with an inspection area, and performing a predetermined process as described below.

Now, it is assumed that there is a defective portion A and a periodic unevenness portion B on the detection line 1 of the sample 6 to be inspected as shown in FIG. In such a detection line 1, first, when the sample 6 to be inspected is slightly moved, image data as shown in FIG. Therefore, when these signals are subtracted, a signal shown in FIG. 4B is obtained. Further, when this signal is differentiated, a signal such as a defect or periodic unevenness excluding low frequency components is obtained as shown in FIG. The conventional device described above attempts to detect a defect by simply setting a threshold value with respect to this signal. As a result, as shown in FIG. There is a problem that a large number of uneven portions B are also detected.

In the present invention, unlike the conventional processing as described above, the absolute value of the signal shown in FIG. 3C is obtained from the signal shown in FIG. Then, from the absolute value signal, the magnitude of the change of the pixel data in the short period in the neighboring area as shown by the following is obtained by the smoothing filter processing, and the threshold value distribution as shown by the addition by adding the predetermined value α is obtained. The threshold distribution generated in this manner has a large value in a portion where image data fluctuates due to period unevenness and the like, and has a small value in a portion where the fluctuation is small, as is apparent from FIG.
On the other hand, the effect of independent minute defects is significantly reduced by smoothing. For this reason, when the threshold distribution shown in the comparison process is compared with the absolute value signal shown in the diagram, the defective portion can be detected without detecting the fluctuation of the short period as shown in the diagram.

It should be noted that the present invention is not limited to the above embodiment, and various modifications are possible. In the above embodiment, the transmission illumination unit is used for the optical system for obtaining the image data of the sample to be inspected. However, if the optical system is for obtaining the defect of the periodic pattern of the industrial product as the image data, the sample to be inspected is illuminated. Then, an optical system for capturing an image using the scattered light or another optical system may be employed. In addition, two samples of image data are captured by slightly moving the sample to be inspected, these are subtracted, and further differentiated. However, the size of the sample is determined as shown in FIG. A different process may be performed as long as the process can obtain a fluctuation signal having the following. For example, a threshold distribution signal such as a rectification of a signal of a high frequency component corresponding to a defect or a periodic unevenness of a sample to be inspected or a variation signal corresponding to a defect or a periodic unevenness of the sample to be inspected. Should be obtained.

〔The invention's effect〕

As is apparent from the above description, according to the present invention, a signal that has detected a changed portion is subjected to a smoothing filter process,
Since the threshold distribution is obtained based on this signal and the comparison process is performed, a pixel showing a larger change with respect to a change in surrounding image data can be detected as a defect. Therefore, in addition to correcting the sensitivity distribution due to shading, resolution distribution, etc., it is possible to prevent the unevenness of the product from being detected as a defect.
Stable and highly sensitive inspection can be performed.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining one embodiment of a defect detection method according to the present invention, and FIG. 2 is a block diagram showing one example of an apparatus for implementing the defect detection method according to the present invention. 1 camera 2 image processing unit 3 control unit 4
... display, 5 ... drive mechanism, 6 ... sample to be inspected,
7 ... Diffusion plate, 8 ... Lamp, 9 ... Power supply, 10 ... XY stage.

Claims (1)

(57) [Claims] 1. A defect inspection method for inspecting a defect in a periodic pattern of a sample to be inspected, wherein a variation signal caused by a defect or period unevenness is extracted from image data obtained by photographing an inspection area, and the variation signal is obtained. A defect inspection method comprising: obtaining a detection threshold distribution by applying a predetermined threshold to a signal by performing a smoothing filter process; and detecting a defect by comparing the detection threshold distribution with the variation signal.