JPH0332722B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a surface defect extraction device for extracting defects on a fine pattern having a lattice portion, such as a color filter.

Defects of color filters having a lattice portion that are bonded directly to a solid-state image pickup plate for colorizing the solid-state image pickup plate include adhesion of foreign matter and scratches. Figure 1 shows an enlarged view of the main parts of the color filter. Fig. 2 shows an enlarged view of the main parts of the same color filter with the main foreign objects and scratches.

If there are foreign objects or scratches on the color filter, part of the solid-state image pickup plate may be damaged when it is bonded to the solid-state image pickup plate, or an adverse effect such as color unevenness may occur on the image. For this reason, conventionally, inspectors thoroughly inspected the surface condition of the color filter by enlarging it with a microscope and visually observing it.

In addition, we have traditionally cut out two-dimensional binary images,
There is also a method of detecting surface defects using a scaling method, but although this method can extract minute defects, it is difficult to extract defects when foreign objects or scratches are large. In addition, a reference grid pattern is input into a large computer, and the grid information is extracted by matching this reference pattern with the grid information of the object to be inspected. Processing is complicated because defects other than grid information are extracted by pattern matching with the digitized video signal. Third
The configuration will be explained using figures.

In the figure, 1 is the object to be inspected, 2 is the object to be inspected 1
3 is a circuit that converts the photoelectrically converted signal into a binary value. 4 is a circuit that stores the binarized signal. The binarized signal of the inspected object 1 stored in the same configuration is compared with the reference information of the inspected object 1 in a defect-free state that has been inputted in advance to a large-sized computer, and the difference between the two is determined.

In this way, defects in the object to be inspected 1 are extracted. However, with such a configuration, the apparatus becomes large-sized and the processing time is long, so it is not very practical to introduce it into an actual production process. For this reason, visual inspection is commonly performed at present.

However, defect inspection by visual inspection by an inspector has the disadvantage that since the defects themselves are minute, it is difficult to obtain stable inspection accuracy due to detection errors, and the inspection time is extremely long.

The present invention provides a surface defect extraction device that extracts defects other than the lattice portion of an object to be inspected having such a lattice portion with a simple configuration.

The configuration of the surface defect extraction device of the present invention will be explained with reference to FIG. 6 is an object to be inspected such as a color filter having a grating section; 7 is a device such as a television camera that magnifies the object to be inspected 6 with a microscope and converts it photoelectrically by scanning the optical image; 9 is a device that converts the signal into electricity. This is a binarization circuit that performs binarization. FIG. 5 shows an enlarged view of a portion of the binarized inspection object 6. In FIG. 5, b, c, and d indicate foreign objects and scratches. e
is the grid line part. There are two main ways to extract defects such as foreign matter and scratches on color filters. One is the extraction of foreign matter that becomes an obstacle when pasting color filters onto a solid-state image sensor. Targets large items. The other is the extraction of defects such as foreign objects and scratches that are an obstacle on the image of the solid-state image pickup plate. Even if objects smaller than the grid width are present in the target grid section, they do not become an obstacle and do not need to be extracted.

Therefore, in order to extract the size, position, and number of defects in the inspection object 6, the binary signal 18 is processed in two paths. One is a path for extracting foreign substances larger than the grid line width. The other is the object to be inspected 6-2
This is a path for extracting grid information from the digitized signal 18 and extracting foreign objects and scratches that may be obstacles on the image of the solid-state imaging board.

First, a description will be given of defect extraction targeting the entire color filter for foreign particles larger than the grid line width. The object to be inspected 6 obtained through the binarization circuit 9
The binarized video signal 18 is sent to the image serial/parallel circuit 1
0, a signal is extracted two-dimensionally, and the matching circuit 11 extracts information on foreign objects larger than the grid line width in time series.

The image serial/parallel circuit 10 converts the binarized signal 18 of the inspected object 6 extracted from the binarization circuit in time series to the inspected object 6 using a serial/parallel converter 21 such as a shift register shown in FIG. This is a circuit that uses picture elements to obtain signals two-dimensionally.

The matching circuit 11 converts the time-series two-dimensional signal 19 extracted by the image serial/parallel circuit 10 into an AND-configured circuit as shown in FIG. This is a circuit that inverts its logic state when A can have any shape as long as it is larger than the vertical and horizontal grid line widths.

The output signal 20 of the matching circuit 11 is input to the defect extraction circuit 12 shown in FIG. Defect extraction circuit 12
Then, defect signals are extracted from the output signal 20 of the matching circuit 11, the number and size are counted, and the CPU
Defect information is stored in 16 RAMs (Random Access Memory).

This content allows CPU (Central Processing)
Unit) 16 recognizes foreign objects larger than the grid line width. FIG. 5 is an enlarged view of a part of the object to be inspected 6b,
C and d indicate foreign matter and scratches, but when they pass through the image serial/parallel circuit 10 and the matching circuit 11, they become as shown in FIG. 8, and foreign matter and scratches smaller than foreign matter A in FIG. 7 are not extracted. Therefore, foreign matter and scratches b and c are extracted, but foreign matter d is not extracted.

Next, a system for eliminating grid information from the binary signal 18 of the inspected object 6 and extracting defects such as foreign objects and scratches will be explained.

The binarized signal 18 enters the image serial/parallel circuit 10 and also enters the storage circuit 13. In the memory circuit 13, the grid information is erased by the operation of the grid erase circuit 14.

The lattice erasing circuit 14 takes out the signal for one scan of the object to be inspected 6 from the memory circuit 13, matches the reference lattice information with the lattice information of the object 6 to be inspected, and sequentially updates the scanning to erase the entire lattice portion. to erase. Fifth
When the inspected object 6 shown in the figure passes through this path, it becomes as shown in FIG. 9. In other words, foreign matter, scratch b,
c and d are all extracted. However, after extracting the lattice information, all pixels are extracted, so the foreign object d, which was smaller than the measurement area A shown in FIG. 7 and was not extracted, is extracted. Foreign matter b on the grid due to erasing
is divided into four parts.

Next, the defect information extracted after passing through each path is grasped by the defect extraction circuit 12 and defect extraction circuit 15 in terms of number and size, and the CPU 16
Defect information is stored in RAM. The CPU 16 integrates the information and obtains the foreign matter and scratches b, c, and d in a clear form as an extraction result, as shown in FIG.

While performing integrated treatment of foreign objects and scratches in this way,
After the defect extraction circuit 12 and the defect extraction circuit 15 operate, the CPU 16 operates the drive circuit 17.
The mounting table 8 for the object to be inspected 6 is moved, and defect extraction for the next screen is executed. This process is repeated to extract defects in the entire area of the object 6 to be inspected.

The surface defect extraction device of the present invention obtains the following effects by adopting the above configuration.

First, visual inspection can be automated, saving labor and improving inspection accuracy. Furthermore, foreign objects and scratches on patterns such as color filters with grid portions can be extracted in a short time with a simple circuit configuration without the need for complicated pattern matching using a large computer.

Furthermore, after erasing the extraction part of foreign matter and scratches larger than the grid line width and the grid, the foreign matter and scratch extraction part of each pixel are processed separately to eliminate foreign matter and scratches that exist across the grid. can be easily recognized.

As described above, the surface defect extraction device of the present invention extracts defects of an object to be inspected having a lattice portion with a simple configuration, and has high industrial utility value.

[Brief explanation of drawings]

Fig. 1 is an enlarged view of the main part of the object to be inspected, Fig. 2 is an enlarged view of the main part showing the state in which foreign matter is attached to the object to be inspected, and Fig. 3 is a diagram showing the configuration of a conventional surface defect extraction device.
Fig. 4 is a block diagram showing the configuration of a surface defect extraction device according to an embodiment of the present invention, Fig. 5 is a drawing showing a state in which an object to be inspected has foreign matter and scratches, and Fig. 6 shows a foreign matter larger than the grid line width. FIG. 7 is a diagram for explaining the size of foreign matter extracted; FIG. 8 is a diagram showing the result of extracting foreign matter and scratches larger than the grid line width; FIG. 9
The figure shows the result of extracting foreign objects and scratches from all picture elements after erasing the grid, and Figure 10 shows the result of integrating the extraction results shown in Figures 8 and 9 through arithmetic processing on the CPU. It is a diagram. 6...Object to be inspected, 7...Photoelectric conversion device, 9...
Binarization circuit, 10... Image serial/parallel circuit, 11...
Matching circuit, 12... Defect extraction circuit, 13...
...memory circuit, 14...grid erasure circuit, 15...defect extraction circuit.

Claims (1)

[Claims] 1 means for photoelectrically converting an object to be inspected having a grating portion while scanning it; a binarizing means for binarizing the electrical signal from the photoelectric converter at a predetermined level; a first defect extraction means that extracts the signal of the binarization means and determines that the two-dimensional area is defective when the logical state is a predetermined value; the first and second defect extracting means, comprising: a lattice part extracting means for comparing and extracting a lattice part signal; and a second defect extracting means for detecting a defect depending on the logic state of the output of the lattice part extracting means; 1. A surface defect extraction apparatus, characterized in that the defect extraction signals extracted from the above are integrated to extract a defective part of the object to be inspected.