JP2718510B2 - Inspection method of colored periodic pattern - Google Patents

Description

The present invention relates to a colored unit such as a color separation filter used for an image pickup tube used for a color television camera or a solid-state image pickup tube element, and a color filter used for a liquid crystal display device. The present invention relates to a method of inspecting a periodic pattern in which turns are repeatedly arranged for defects, unevenness, and the like.

[Conventional technology]

Conventionally, CCD color filters and liquid crystal color filters
Defects or irregularities in the pattern will result in defects in the image taken or displayed, and if there are minute irregularities on the surface, the product using the color filter will be defective. After the inspection,
Defective color filters must be excluded.
Conventionally, such color filter inspection has been performed by a microscope inspection and a visual inspection using various illuminations, but there have been problems such as individual differences, missing defects, and the like.

Therefore, various methods have been proposed to solve these problems.

An inspection method in which a periodic pattern is photographed by a television camera and subjected to image processing will be described with reference to FIGS.

In the inspection apparatus shown in FIG. 8, an inspection object 46 having a periodic opening as a unit pattern 51 as shown in FIG.
In order to detect an abnormality in the opening area of the inspection object 46, the inspection object 46 is illuminated by a transmission illuminating unit constituted by an incandescent lamp 48 illuminated by a DC power supply 49 and a diffusion plate 47, and an inspection area is photographed by a TV camera 41. The image processing device 42 A / D converts the output signal of the TV camera into digital image data, and performs various image processing including addition and subtraction of screens at high speed by a frame memory and an arithmetic unit. The control device 43 controls the image processing device 42 and the pattern moving mechanism including the XY stage 50 and the driving mechanism 45 to move the pattern. In FIG. 9, 52,
53 is a unit pattern having a defect.

The imaging conditions under which the change due to the unit aperture of the video signal by the TV camera 41 is negligible, for example, the pattern area corresponding to one pixel is such that about ten unit apertures 11 are included, and the direction of moving and displacing the pattern is the scanning line of the TV camera 41 The case where the displacement distance of the pattern in the direction is an integral multiple of the pixel pitch will be described with reference to FIG.

The light transmittance distribution on a straight line passing through the place where the pattern has a defect is, for example, as shown in FIG.
The defect 54 whose opening area is larger than that of the normal pattern 51 as shown by, that is, a change 54 in light transmittance due to a white defect, and the defect shown in FIG.
As shown by 52, a defect having a smaller opening area than the normal pattern 51, that is, a change 55 in light transmittance due to a black defect is detected. FIG. 10 (b) shows a video signal scanned on the same line as that in FIG. 10 (a), and a gradual signal change (shading) due to illumination unevenness of the pattern, sensitivity unevenness of the imaging surface, and the like. ) And random noise generated in the video signal processing device, and local changes 56 of the signal due to dust or the like attached to the optical system.

By adding a plurality of frames to such a video signal, when the number of additions is N, the ratio of the random noise component is (FIG. 10 (c)). next,
When the same screen addition processing is performed by displacing the pattern,
As shown in FIG. 10 (d), the signal due to a defect on the pattern is moving along with the movement of the pattern, but the position of the signal 56 due to shading of the imaging system, dust of the optical system, etc. is not displaced. Therefore, when the data shown in FIG. 10 (d) is subtracted from the data shown in FIG. 10 (c), the signal 56 due to shading or dust contained in the quantity data is erased, and the signal 56 due to the change in the light transmittance of the pattern is removed. Only the reduced random noise component remains, and as a result, the signal due to the defect appears at the position separated by the number of pixels according to the amount of movement of the pattern and the value difference from the average value in the vicinity is almost the same, and the sign is inverted. The order of reversal is reversed depending on the type of defect (white defect, black defect).

In the image data processed as described above, the brightness of only the defective portion is locally changed, so that it can be easily recognized as a defect when observed on a monitor. The types of defects can be identified in this order.

Next, a method of moving a pattern will be described with reference to FIG.

As shown in FIG. 11, there is the object to be inspected located P ₀ to P _8, the movement of the aforementioned corresponding to P ₀ and P ₁ in FIG. In this case, when processing is performed only with image data obtained from two locations, a spot caused by a change in the aperture ratio in the H direction in the figure is detected, but a change in another direction, for example, the V direction is detected. There is a defect that the detection sensitivity is anisotropic that a large spot with a small change in the H direction is not detected. For example, P ₀ ,
Based on the image data obtained by imaging at P ₁ , P ₂ , P ₃ , and P ₄ , P ₀
And P _1, P ₀ and P _2, P ₀ and P _3, it is possible to eliminate the anisotropy by performing predetermined processing for each of a combination of P ₀ and P _4. Further, even if the detected based on image data obtained by subtracting the sum of the image data from the image data of the center position P ₀ at each position which are arranged on the circumference as P ₈ from P ₁ in FIG. Similarly, anisotropy can be avoided, and a value approximating the curvature of the brightness distribution that is easily visually responsive is obtained, and the inspection result is close to a visual inspection. The image data obtained by imaging at each position includes the above-described random noise component, shading component, and fixed noise component. The random noise component is obtained by adding a plurality of frames at each moving position. Can be suppressed by
Further, the shading component and the fixed noise component can be eliminated by performing the operation of the image data so that the total number of frames of the image data becomes “0”. For example, 32 frames at a position P ₀ of FIG. 11, performs addition of 8 frames in P _1, is subtracted image data in P ₀ image data from the quadruple image data at P ₁ both The sum of image data is "0"
And the jading component and the fixed noise component are eliminated. Further, at each position P ₈ from P _1, when performing the screen addition of each four frames, since the addition result of the image data P ₈ from P ₁ corresponds to the sum of the image data of 32 frames P _By subtracting from the result of adding the image data for 32 frames at the position of ₀ , the shading component and the fixed noise component are similarly eliminated, and the brightness distribution
A dimensional derivative is obtained. Furthermore, when smoothing is applied to the image data in which the shading component and the fixed noise component are reduced by the above processing, the random noise component is further reduced, and it becomes possible to detect an extremely slight uneven component. It is also possible to suppress a change in image data due to a minute defect or short-period irregularity.

A method has also been proposed for automatically determining whether a product is good or bad based on image data that has been subjected to image processing as described above.

FIG. 12 shows an example of measurement under the same conditions as in FIG. 10, and FIG.
12 (a) to 12 (e) are the same as FIG. 10, where A is a portion where the aperture ratio changes slowly, B is a portion where the aperture ratio changes periodically, and C is a portion where the aperture ratio changes. A large and isolated portion, D, is a local change component (fixed noise component) of the video signal due to contamination of the optical system. FIG. 12 (e) shows that components due to the change in the aperture ratio of the test object are extracted. In this case, the moving amount of the test object differs depending on the state of the unevenness to be detected. However, the moving amount of the test object increases with an increase in the cycle of the change in the unevenness, and is converted into the number of pixels of the frame memory. From about 2 pixels to about 20 pixels. Next, when the average value of the image data of a sufficiently large area is subtracted from the image data of FIG. 12 (e), the result becomes as shown in FIG. 12 (f), and the image data of FIG. Is differentiated from that shown in FIG. 12 (g), and a predetermined threshold value is set according to the nature of the unevenness to be detected.
Unevenness can be automatically detected by setting S ₁ and S ₂ . FIG. 12 (h) sets threshold values S ₁ and S ₂ for the image data of FIG. 12 (g), and shows “1” when the threshold value is exceeded and “0” when the threshold value is not exceeded. It is quantified data. Then, to detect the isolated unevenness as shown in C of Figure 12 (a), the product if a defect, the threshold such as the S ₁ to the image data of Fig. 12 (f) or (g) Should be set. In the case of detecting a periodically changing unevenness as shown by B in FIG. 12 (a) and determining that the product is defective, the image data of FIG. 12 (f) or (g) is
A threshold value such as S2 is set, and ₂ is set as shown in FIG.
After the conversion into the coded data, the neighboring pixels are added to convert the data into the number of unevenness in a predetermined area as shown in FIG. 12 (i), that is, the density data. _Three
By setting and comparing, it is possible to detect only the periodically changing unevenness, and determine the quality of the product from the result.

[Problems to be solved by the invention]

However, when the above-described inspection method is applied to the inspection of a multi-color periodic pattern, 1-colored R, G, B, etc.
In order to treat a set of patterns as a unit pattern, as shown in FIG. 13, the arrangement pitch of the periodic pattern is increased and the period with the imaging pixel approaches, so that moiré of the captured image becomes strong, and the detection capability is high. There is a problem that the signal level of each color of the video signal is made uniform and the pitch of the pattern in the imaged signal is reduced to prevent moire. Abnormal shape of the boundary of each color like 21, 22
Cannot be detected.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and an object of the present invention is to provide a method of inspecting a colored periodic pattern which can reliably inspect a color filter or the like with high sensitivity.

[Means for solving the problem]

For this purpose, the present invention provides a method of inspecting a colored periodic pattern from image data obtained by irradiating a sample having a colored periodic pattern with light and capturing an image of the sample. By a plurality of imaging means for imaging, R,
A video signal of a single color image of each of G and B is obtained, and R,
G, B monochromatic video signals and their R, G, B
The video signal of the single color image is synthesized so that the video signal level corresponding to each color is aligned, and the video signal of the synthesized image is subjected to defect detection processing.

[Action]

The present invention can detect an abnormality in the shape of the boundary of each color of the colored periodic pattern by reading the image of the wavelength component corresponding to each color constituting the colored periodic pattern, and By making the video signal levels uniform, the pitch at the signal level can be shortened to prevent the occurrence of moire, and the resolution can be set high to perform high-sensitivity defect detection.

〔Example〕

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

FIG. 1 is a diagram showing an embodiment of the present invention, and FIG. 2 is a diagram showing an embodiment of the configuration of a light source. The same reference numerals as in FIG. 9 denote the same contents. 1a to 1c are R, G, and B light sources, 2a to 2c are optical fibers, 3 is an optical fiber bundle, 4 is an exit port, 5 is a lens, 6a to 6b is a camera control unit (CCU), and 7 is The synthesis circuits 8a to 8d are frame integration circuits.

The R, G, and B light sources 1a to 1c collect light from a halogen lamp 11 with a reflecting mirror 12 and pass light of a specific color through a color filter 13 with optical fibers 2a to 2c as shown in FIG. It is configured to receive light, and the light intensity can be adjusted independently of each other. Optical fibers 2a to 2c receiving light from each light source
The optical fiber bundle 3 is bundled so that the fibers of the respective colors are uniformly mixed at the exit 4. The sample 46 placed on the XY stage 50 is illuminated using the fiber as a light source, and an image is taken by a three-tube or three-plate color television camera 41. The color television camera 41 separates an optical image formed by a photographing lens into R, G, and B lights by optical means such as a color separation prism, and reads each image with a separate imaging tube or CCD. R, G, and B video signals obtained from the color television camera 41 and a signal obtained by combining them (equivalent to a black and white image) are A / D converted by a frame integration circuit.
Integration processing is performed using the frame memory. Image processing device
At 42, four types of image data obtained by a series of imaging operations are sequentially processed to detect defects and unevenness. That is, RG
When the spectrum of the illumination light is set by adjusting the light amounts of R, G, and B so that the signal level becomes the same for each color while observing the B composite image, the arrangement pitch of the color filters becomes R,
What used to be G and B as a unit, as shown in FIG.
As a result, the occurrence of moiré is suppressed, the resolution is set high, and a defect requiring a high detection capability such as minute black can be detected, and a highly sensitive defect can be detected. In addition, a defect or unevenness that cannot obtain a contrast unless it is colored light as shown in FIG. 4 or shows a strong contrast with monochromatic light can be detected from a monochromatic image such as R, G, and B.

The imaging of each of the R, G, and B wavelength components can be performed by either serial processing or parallel processing. However, if the imaging of each wavelength component is performed simultaneously, the inspection time can be reduced to 1/3 of the serial processing. It can be shortened.

FIG. 5 is a view showing another embodiment of the present invention in which a cooled CCD camera is used as an imaging device. In the drawing, reference numeral 61 denotes a cooled CCD camera.
A camera, 62 is an image processing device, 64 is a shutter, 65 is a shutter driving device, and 66 is a sample.

The cooled CCD camera 61 is a CCD camera that can be cooled for a long time in a dark area by cooling it by an electronic cooling method and reducing dark current and noise to a negligible level. It is characterized by good image quality, and can clearly show dark areas that could not be projected with conventional high-sensitivity television cameras with high image quality.
It is possible to detect up to several photons per pixel per second.

Using such a cooled CCD camera 61, the light amount of each color is adjusted so that the signal level of each color becomes the same when the image is taken, and the sample 66 is irradiated with the optical fiber 3 and imaged by the CCD camera. If the image data taken without the sample is I ₁ , the image data taken with the sample in is I, and the image data taken with the shutter 4 closed is I ₀ , the transmittance T of a point on the sample is Can be calculated as Here, I, I ₀ , and I ₁ are the image data at the corresponding positions. If the image data when the shutter is closed (light amount = 0) can be ignored, the transmittance can be obtained as T = I / I ₁ . This calculation can be performed by an inter-screen calculation after each image data is stored in the frame memory by the image processing device 2. If the number of pixels of the cooled CCD camera is 512 × 512, about 250,000 transmittance data can be obtained by this calculation. Since the image data thus obtained does not include components such as shading and dust of the imaging system, the image data is subjected to image processing according to each inspection item as described above, so that the image data can be measured in one measurement. An inspection can be performed. The light source can be stably keep the image data I ₁ is stored in the memory if, which it is not necessary to perform measurement for each sample when to use, to shorten the measurement time.

With a normal solid-state image sensor or image pickup tube, the linearity of the light amount and the video signal is not sufficient, and the effect of thermoelectrons is large, making high-precision measurement difficult. Is the spatial resolution, linearity, S / N
Is good, but there is a problem that the imaging time is long (about 1 ms per pixel, about 4 minutes as 250,000 pixels)
In the present invention, by using the cooled CCD camera 1, an image can be captured in about several seconds and the linearity is improved.

Defect detection is performed by differentiating the image data to obtain image data equivalent to the result of conventional imaging (frame integration) → sample movement → imaging (frame integration) without moving the sample. After that, if the same image processing is performed, the defect can be detected. In this case, for example, feature extraction may be performed by using a spatial filter of differential processing as shown in FIGS. 6 (a), 6 (b) and 6 (c). One-dimensional edge extraction can be performed by using the spatial filter shown in FIGS. 6A and 6B, and two-dimensional edge enhancement can be performed by using the spatial filter shown in FIG. 6C. it can. And the variation of the image data,
The spatial filter may be selected in accordance with the resolution characteristics, the nature of the defect to be detected, and the like. The white / black defect is identified by using the spatial filter shown in FIG. However, it can be identified by the size of the target pixel with respect to the surrounding pixels.

Also, by using the spatial filter of FIG. 6 (c) for the detection and determination of the unevenness, the same result as when performing a series of imaging operations as described in FIG. 12 is obtained. Similarly to the above, detection and determination can be performed by performing screen addition processing and threshold setting.

By the way, in the transmittance measuring method shown in FIG. 5, since a variation in data for each pixel directly affects the data, image data of a small area around a measuring point, for example, 5 It is necessary to calculate an average value such as × 5pix to 10 × 10pix. When the number of measurement points is limited, the processing time by the CPU requires less time.However, when the number of measurement points is large, the processing time by the CPU requires a lot of time. After performing a smoothing filter process, which is one type of image processing, on a given frame memory and reading out predetermined image data, the speed can be increased.

As described above, the cooling type CCD is characterized by the linearity of the video signal to the integrated light quantity is good, when measuring low transmittance sample image data I ₁ captured without sample Even if the value is set to a value close to max, the value of the image data I captured with the sample put in becomes small, and a slight error in linearity, a dark current, and the like affect the transmittance value.

For this reason, when imaging I and I ₁ , it is more accurate to image the two image data levels so that the two image data levels are almost equal and have a sufficient size. , A correction operation is required.

As one method for that, the exposure time at the time of I ₁ imaging,
By operating the shutter so as to be 1 / T of I, I and I ₁ can be made substantially the same value.

However, the mechanical shutter has a variation in operation time, and therefore has a disadvantage that an error increases when the shutter opening time is short, and when this error cannot be ignored,
It is sufficient to measure the shutter open time and correct the data based on the measured value.

FIG. 7 is a diagram showing another embodiment of the present invention for realizing this, and the same numbers as those in FIG. 5 indicate the same contents. In the drawing, 70 is a half mirror, 71 is an optical sensor, 72
Is a measuring device.

In the present embodiment, a part of the light at the time of imaging is detected by the sensor 71 by the half mirror 70, and the exposure time is obtained by measuring the time during which the detection signal is obtained by the measuring device 72. Correcting the data by way exposure time obtained, it is possible to substantially equalize the two image data level of I and I _1.

Further, by changing the accumulation time of the cooled CCD, for example, it may be equal to the level of the I and I ₁ by lengthening towards the image data I captured put sample.

Also it can be changed brightness of the light source at the time of imaging of the I and I ₁ obtain similar results.

[Effects of the Invention] As described above, according to the present invention, it is possible to perform high-sensitivity defect detection by setting a high resolution by preventing the occurrence of moire due to a colored pattern, and to perform coloring from an image using monochromatic light. It is possible to reliably detect a defect or unevenness that cannot obtain a contrast or a strong contrast with monochromatic light unless it is colored light such as a pattern shape abnormality.

[Brief description of the drawings]

FIG. 1 is a diagram showing an embodiment of the present invention, FIG. 2 is a diagram showing an embodiment of the configuration of the light source, FIG. 3 is a diagram showing a relationship between a coloring periodic pattern and an image pickup signal level, and FIG. FIG. 5 is a diagram showing an abnormal shape of a colored pattern, FIG. 5 is a diagram showing an embodiment in which an image is taken by a cooled CCD camera, FIG. 6 is a diagram showing a spatial filter, and FIG. FIG. 8 is a view showing another embodiment of the present invention, FIG. 8 is a view for explaining a conventional periodic pattern inspection method, FIG. 9 is a view for explaining a periodic pattern and its defects, FIG. FIG. 11 is a diagram for explaining a conventional detection method, FIG. 11 is a diagram for explaining a pattern moving method, FIG. 12 is a diagram for explaining a method of detecting unevenness, and FIG. FIG. 4 is a diagram illustrating a relationship between signal levels. 1a to 1c: light source for R, G, B, 2a to 2c: optical fiber,
3 ... optical fiber bundle, 4 ... outlet, 5 ... lens, 6a
6c ... CCU, 8a-8d ... frame integration circuit, 41 ... color television camera.

Claims (2)

(57) [Claims] 1. A method for irradiating a sample having a colored periodic pattern with light and inspecting the colored periodic pattern from image data obtained by imaging, capturing images of R, G, and B wavelength components. A plurality of image pickup means obtain a video signal of a single-color image of each of R, G, and B. The video signal of the single-color image of each of R, G, and B and the video signal of the single-color image of R, G, and B are obtained. A method for inspecting a colored periodic pattern, wherein a defect detection process is performed on a video signal of a synthesized image synthesized so that video signal levels corresponding to respective colors are aligned. 2. The method for inspecting a colored periodic pattern according to claim 1, wherein imaging of each wavelength component is performed simultaneously.