JPS62201335A - Speckle inspecting method for periodic pattern - Google Patents

Abstract

PURPOSE:To inspect speckles efficiently with high accuracy by applying a digital image processing method and extracting speckle components. CONSTITUTION:A body 6 to be inspected is lighted by the transmission lighting part consisting of an incandescent lamp 8 powered on by a DC power source 9 and a diffusion plate 7 and an inspection area in the body 6 to be inspected is photographed by a TV camera 1. The obtained video signal is inputted to an image processor 2, wherein the video signal is A/D-converted and stored in a frame memory. Then, the inspection area is moved by the body to be inspected moving mechanism consisting of a driving mechanism 5 and an XY stage 10 to inspect speckles of the object body 6. Thus, speckle components are extracted and displayed on a TV monitor to easily find speckles which are hardly recognized with the eye, thereby inspecting the speckles with high accuracy.

Description

[Detailed Description of the Invention] [Object of the Invention] (Industrial Field of Application) The present invention is applicable to Nyadou masks used in cathode ray tubes for color televisions, color separation filters for color imaging devices, color filters for liquid crystal display panels, and electron tubes. Units with certain optical properties and shapes (hereinafter referred to as unit patterns) such as mesh electrodes, VDT filters, mesh filters for leakage devices, rotary encoders, linear encoders, photomasks for ■C, Funnel lenses, and lenticular lenses, etc. Industrial products or unit patterns that are regularly and repeatedly arranged in one dimension or two dimensions gradually change in their optical properties, shape, and arrangement in one or two dimensions. The present invention relates to a method for inspecting spots caused by non-uniformity in optical properties due to the shape, size, position, etc. of unit patterns of industrial products that are repeatedly arranged.

(Prior Art) Conventionally, the inspection of industrial products having the above-mentioned periodic pattern generally involves measuring elements such as the shape, size, arrangement pitch, or color of the unit patterns that make up the periodic pattern using a special measuring instrument. However, even if the measurement results satisfy the predetermined quality standards, if the optical properties of the entire product, in which a large number of unit patterns are arranged, are significantly non-uniform, the product must be removed. It becomes necessary to remove it as a defective product.

For example, when inspecting Nyadou masks used in color television cathode ray tubes and color separation filters for color imaging devices, the above-mentioned elements are inspected for each unit pattern using a special measuring instrument, and furthermore, the final product is inspected by artificial means ( Problems to be Solved by the Invention) According to the conventional method, in a periodic pattern consisting of a repeating arrangement of unit patterns, the step of inspecting spots on an industrial product having the unit pattern is performed with the naked eye or using a microscope. However, since it is a sensory test, it requires a lot of manpower to test a large number of products, and because it is a sensory test, it lacks reliability and accuracy. .

Therefore, an object of the present invention is to provide a method that makes it possible to efficiently and accurately inspect the above-mentioned industrial products for spots.

[41°4 structure of the invention] (Means for solving the problem) The present invention provides a periodic pattern consisting of a repeating arrangement of unit patterns, the shape, size, arrangement pitch of the unit pattern,
Alternatively, in a method for automatically inspecting spots caused by non-uniformity of color, etc., the cycle 1 raw pattern is divided into a plurality of pixels by an image input device, and digital data for each pixel is stored in the image memory in units of frames. means for storing image data in a frame 1'1, and random noise components generated in the step of obtaining the image data, aging components caused by the brightness distribution of the light source, and adhering to the optical system or the product. A means for suppressing pulsed noise (hereinafter referred to as fixed noise) caused by dirt to make it easier to recognize by extracting mottled components of the periodic pattern, The above object has been achieved by providing a means for determining whether a product is good or bad.

(Function) According to the present invention described above, by extracting the mottled component of a periodic pattern by applying a digital image processing method, the mottled component can be highlighted and displayed, making it easier for visual inspection. In the process, when used as an auxiliary means for visual recognition, spots that are difficult to catch using conventional methods can be easily recognized.

Furthermore, by comparing the extracted spot components with the judgment criteria using electrical means, it is possible to automate the spot inspection, which eliminates the need for a large amount of manpower and can be performed quickly in accordance with the flow of the manufacturing process. Another effect is that highly accurate inspection can be performed.

(Example) Hereinafter, the present invention will be described in more detail based on an illustrative example.

FIG. 1 is a block diagram showing an example of an apparatus for carrying out the method according to the present invention, and numeral 1 in the figure is a TV camera.

2 is an image processing device δ, 3 is a control unit, 4 is an image display display, 5 is a drive mechanism, and 6 is an object to be inspected.

7 is a diffuser plate, 8 is an incandescent lamp, and 9 is a DC power supply.

10 indicates an XY stage. To explain the operation of the present invention, first, the object to be inspected is illuminated by a transmitted illumination section consisting of an incandescent lamp 8 and a diffuser plate 7, which are lit by a DC power source 9.
The video signal obtained by photographing the inspection area in the test body 6 with the TV left camera is input to the image processing device 2, and in the image processing device, the video signal is Convert it to D, store it in the frame memory, and perform the processing described below.''1
, drive a 4'f/i 5 under the control of the control device 3.
The object to be inspected 6 is moved through the inspection area by an object moving mechanism including an XY stage 10 and an XY stage 10, and the object 6 to be inspected is inspected for spots.

FIG. 2 shows an example of the object to be inspected 6, which is a periodic pattern having periodic openings as unit patterns 11.

FIG. 3 schematically shows data corresponding to each processing step as an analog waveform in order to explain the processing contents of the image processing device 2 and its effects.

FIG. 3 (al) is an example of the aperture ratio distribution of the object to be inspected 6. Imaging conditions are such that the shape of the unit pattern 11 is ignored in the video signal waveform of the TV left camera, for example, the frame memory of the image processing device 2 The figure shows the case where the number of unit patterns 11 on the object to be inspected 6 corresponding to one pixel is about 10. In the figure, A shows a part where the aperture ratio changes gradually, and B shows a part where the aperture ratio changes slowly. The periodically changing portion C indicates a portion where the aperture ratio changes largely and is isolated.

FIG. 3(b) shows the video signal waveform for one scanning line of the TV left camera input to the image processing device 2.

This signal component corresponds to the aperture ratio distribution shown in FIG. A random noise component generated in the video signal processing circuit, and a local variation component (fixed noise component) D of the video signal due to contamination of the optical system, etc. are shown.

□ Figure 3 (C1 shows the image data as a result of suppressing the random noise component by the image processing device 2,
By storing the video signal in a frame memory and adding N frames, the signal component that does not change between frames increases, whereas the random noise component has no correlation between frames, so the average amplitude increases. Random noise components are suppressed using attenuation to 1/what. Note that N frame memories may be used, but 1 frame memory may be used.
A similar effect can be obtained by using two frame memories in a cyclic configuration.

Further, in FIG. 3 (dl is after slightly moving the inspected object 6,
This figure shows the results of suppressing the random noise component described above.

FIG. 3 (tel) shows the image data obtained by subtracting the image data shown in FIG. 3 ((1) from the image data shown in FIG. 3 (Cl), and FIG. 3 (C).

(The components that do not change due to minute movements of the object 6 to be inspected, such as the aging components and fixed noise components included in the image data of dl, are eliminated, and the component due to the change in the aperture ratio of the object 6 to be inspected is reduced to 1. The movement of the subject 6 varies depending on the condition of the spot to be detected, but as the period of variation of the spot increases, the minute movement of the subject 6 increases. The amount of movement also increases, and in this embodiment, a suitable value ranges from 2 pixels to 200 pixels in terms of the number of pixels in the frame memory.

The above is a process based on the image data obtained by moving the inspected object 6 once in one dimension and capturing images before and after that.
This shows a method for extracting mottled components.

FIG. 4 schematically shows the movement positions in order to explain the method of moving the inspected object 6 minutely. In the figure, Po to Ps correspond to the imaging position of the inspected object 6. There is. For example, the position imaged by the above movement is P in the same figure.
. and corresponds to Pl. However, when processing is performed using only image data obtained from these two locations, spots caused by changes in the aperture rate in the H direction in the figure can be detected, but in other directions, for example, in the V direction. It has the disadvantage of causing anisotropy in detection sensitivity in that spots with large changes and small changes in the H direction are not detected.
``2+ PK* P4 Based on the image data obtained by capturing the image, Po and Pl, Po and I'2. PG,...!:
Figure 3 shows each rough combination of Ps, PQ and P4 (
The anisotropy is eliminated by performing the process described in el. Also, the sum of the image data at each position arranged on the circumference from Pl to Ps in Fig. 4 is calculated as the center position Po.
Even if the detection process is performed based on the image data obtained by subtraction from the image data of It has the characteristic of producing inspection results close to visual inspection.

By the way, the above explanation has focused on the method of moving the subject 6, but the image data obtained by imaging at each position includes the aforementioned random noise component, aging component, fixed noise component, etc. The following describes how to reduce these unnecessary components. First, random noise components can be suppressed by adding up a plurality of frames at each movement position shown in FIG. 4 in the same manner as described above. Furthermore, the aging component and the fixed noise component can be eliminated by performing calculations on the image data so that the sum of the number of frames of the image data becomes "0". For example, add 32 frames at the position of Po in FIG. 4 and 8 frames at Pl, and add the image data of 4 times the image data at Pl to P. If you subtract the image data of 1 image, both image data corresponds to 32 frames.) Since the image data is added, the sum of the number of frames of both image data becomes 0, which is the aging component. and fixed noise components are eliminated. To give yet another example, the fourth
If screen addition for 4 frames is performed at each position from Pl to Ps in the figure, the addition result of image data from P to Ps corresponds to the addition result of image data for 32 frames. By subtracting from the result of adding 32 frames worth of image data at the position of Po, the aging component and fixed noise component are eliminated as in the above example, and the second derivative of the brightness distribution is obtained. can get.

Furthermore, by applying smoothing processing to the image data whose aging components and fixed noise components have been reduced through the processing described in step 1 or above, the random noise components are further reduced, making it possible to detect extremely slight mottled components. It is also possible to suppress changes in image data due to minute defects and short-cycle spots, and to select and detect long-cycle spot components.

The above-described processing is performed on the entire area of the object 6 to be inspected, and the obtained image data, in which the entire area of the object 6 to be inspected is processed, is displayed on a TV monitor, thereby detecting the occurrence of spots on the object 6 to be inspected. The spots appear as changes in the brightness level only in the areas where they are present, so spots can be easily recognized even during visual inspection.

Next, a method for automatically determining whether a product is good or bad will be described based on the image data that has been subjected to the above image processing.

Figures 3 (f) and (g) show image data obtained by removing the DC component from the image data in Figure 3 (e), and the artist shown in Figure 3 (fl is Figure 3 1e). The image data obtained by subtracting the average value of the image data of a sufficiently wide area of the data, Figure 3 (gl indicates the image data obtained by differentiating the image data of Figure 3 (el). By setting predetermined threshold values (S1, S2) according to the nature of the spots, the spots are automatically exposed.

Figure 3 (11) shows that the image data in Figure 3 (gl) is set to 7t, thresholds S1 and S2 are set, and those that exceed the threshold are
This is binarized data in which "0" indicates that the value does not exceed "1". To further explain the relationship between threshold 1 and spots, first of all, when detecting an isolated spot as shown in C in Figure 3 (al) and determining it as a product defect, as shown in Figure 3 (fl, (gl) It is sufficient to set a threshold value such as Sl for the data.As another example, when detecting synchronously changing spots as shown in Fig. 3 (B in alO) and determining the product as defective, After setting a threshold value such as S2 for the image data in Figures 3 (r) and (g), it is converted to binarized data as shown in Figure 3 (hl), and further as shown in Figure 3 (il). The binarized data is converted into the number of spots in a predetermined area, that is, density data, and a predetermined threshold fi (ES
By providing s and performing the comparison, only periodically changing spots can be detected, and based on the results, it is possible to determine whether the product is good or bad, and inspection results similar to those obtained by visual inspection can be obtained.

Next, a row will be shown in which spots generated due to periodic changes in the aperture ratio of the object to be inspected 6 according to the above embodiment are removed. The diameter of the unit pattern 1 of the object to be inspected 6 is 100 μm.
A periodic pattern with an arrangement pitch of 300 μm and an arrangement as shown in Fig. 2 is illuminated with transmitted light as shown in Fig. 1, and a TV camera 1 using an image tube is used to obtain GO0.
, 7m x 450ya as the entire inspection area, photo 1 (
1. The total time and image processing time is about 5 seconds, and the change in brightness is 0.5%, which is close to the detection limit of visual inspection, and can accurately determine whether the product is good or bad. We were able to.

Note that as the imaging method and illumination method shown in the present invention, any method can be used as long as a video signal containing information on spots can be obtained. For example, the imaging method is a 1゛V camera using an image pickup tube or a solid-state image sensor (area sensor or line sensor).

Methods using imaging devices using image detectors and flying spot tubes, X-ray TV imaging devices, electronic microscope imaging devices, and i! l: (The bright method is transmitted light illumination,
Methods using transmitted dark-field illumination, reflected dark-field illumination, specular reflection illumination, etc. are mentioned. Furthermore, the properties of the illumination light are as follows:
Coherence, spectral characteristics, polarization, etc. may all be selected so that the SN ratio of the signal when imaging the spot to be detected is high.

〔Effect of the invention〕

As described above, according to the present invention, delicate spots on various industrial products having periodic patterns are photographed with a photographic field of view in which unit patterns constituting the periodic pattern cannot be resolved, and the image data is subjected to image processing. By extracting the mottling component and displaying the mottled component on the TV monitor, it is possible to capture the mottling that is difficult to visually recognize at each station.The characteristics of the mottling can also be determined, and the quality of the product can be determined based on the results.・It is also possible to automatically determine defects, reducing inspection time to 111t.
Significant effects such as improved performance, reliability, and production efficiency were seen.

[Brief explanation of drawings]

FIG. 1 shows steps 1 for carrying out the method according to the present invention.
...'rv camera 2...Image processing device 3...111 details 4...i" V monitor 5... ....... Drive rock structure 6 ..... Test object 7 ..... Diffusion plate 8 ..... Incandescent lamp 9 ...・・・・・・DC power supply

Claims (6)

[Claims] (1) In a method for inspecting spots caused by disturbances in the uniformity of optical properties in a periodic pattern consisting of a repeating arrangement of unit patterns, the periodic pattern is imaged;
A means for decomposing pixels to obtain frame-by-frame image data and a means for moving the periodic pattern to a predetermined position are provided, and the periodicity is determined based on the image data obtained by moving the periodic pattern to a plurality of positions. A method for inspecting spots on a periodic pattern, characterized in that the method is configured to inspect the uniformity of the pattern. (2) The periodic pattern spot inspection method according to claim 1, wherein the image data described above is image data obtained by adding image data of a plurality of frames. (3) A patent claim characterized in that the above image data is image data obtained by processing such that the sum of the number of frames of each image data obtained at each position of the periodic pattern becomes "0" A periodic pattern spot inspection method according to item 1 or 2. (4) The above-described image data is image data obtained by moving and processing the periodic pattern to multiple positions so that the value corresponds to the curvature of the distribution of optical properties in the periodic pattern. A periodic pattern spot inspection method according to claim 1, 2, or 3, characterized in that: (5) The above-mentioned image data is image data obtained by moving and processing the periodic pattern so that it becomes a distribution of the difference in optical properties between two points in a predetermined positional relationship. A periodic pattern spot inspection method according to claim 1, 2, or 3. (6) A patent characterized in that the above-mentioned process is a process in which a portion where the change in the optical properties of the periodic pattern exceeds a predetermined level and the density of pixels exceeds a predetermined value can be recognized as defective. A periodic pattern spot inspection method according to claim 1, 2, 3, 4, or 5.