JPH0981683A - Character symbol inspection device and method therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To judge the prioriety of an inspection character by a high-speed processing so as to match with the judgement standard of a human being. SOLUTION: Inspection images and reference images (MA) composed of image-picked-up multi-valued images are converted to binary images, the templates (20, 21 and 22) of plural different types to be comparison references modeling the image area of an inspection object are prepared from the reference images (MA) converted to the binary images, the defects of blotting, thinning or lacking are discriminated by comparison judgement using the templates (20, 21 and 22) and thus, an inspection is performed with high accuracy. At the time, by considering the line width of the inspection character, propriety judgement close to the judgement standard of the human being is performed. Further, by converting the inspection images to the binary images, extracting the differential area of the inspection images and the templates (20, 21 and 22) and performing the comparison judgement with judgement threshold values (N, H and K), the high-speed processing is performed.

Description

Detailed Description of the Invention

[0001]

[Table of Contents] The present invention will be described in the following order. TECHNICAL FIELD OF THE INVENTION Conventional Technology Problems to be Solved by the Invention Means for Solving the Problems Embodiments of the Invention (FIGS. 1 to 8)

[0002]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character and symbol inspection apparatus and method thereof, and can be applied to, for example, an apparatus for inspecting the shape of a printed character.

[0003]

2. Description of the Related Art Conventionally, the shape inspection of a printed character is performed by capturing an image of the inspection character to be inspected with a television camera, temporarily storing it in an image memory, and comparing it with an image of a master character read from the image memory. . The image of the master character referred to here is an image in which the inspection character used as the reference of the comparative inspection is registered in the master image file.

The master character is registered by taking an image of a character that serves as an inspection reference with a television camera. At this time, the image of the master character is divided into regions for each character according to each typeface of the inspection character. In this way, the comparison inspection is performed for each area divided between the area-divided master character and the area-divided inspection character in the same manner as the master character is divided.

The comparison check is performed based on the normalized correlation between the check character and each corresponding divided area of the master character. That is, if the correlation coefficient between the corresponding areas of the master character and the inspection character is higher than a predetermined value, it is determined as a good product, and if it is less than the predetermined value, it is determined as a defective product.

[0006]

In order to carry out a stable inspection, it is necessary to set the division mode when the inspection character is divided into areas according to the typeface of the inspection character.
However, there is a problem in that it takes a long time to set the division mode. Further, it is necessary to set the judgment level for each of the inspection characters each time, which is troublesome. As described above, the method of setting the parameter for each inspection character takes time and labor, and there is a problem that the inspection efficiency in the inspection process is deteriorated.

[0007] By the way, when the inspection character is missing or bleeding, the line width of the inspection character is changed, and the line width of this character is one of the criteria for judging the defective portion of the inspection character by human sense. . On the other hand, the conventional inspection method has a problem that it is difficult to perform an inspection determination that is close to a human determination standard in consideration of the line width of a character. The present invention has been made in consideration of the above points, and intends to propose a character symbol inspection apparatus and a method therefor capable of performing a high-speed processing of an inspection character and determining whether the inspection character is good or bad so as to meet a human criterion. It is a thing.

[0008]

In order to solve such a problem, in the present invention, an inspection image obtained by picking up and processing an image of an inspected object consisting of printed or displayed characters or symbols is used as a comparison reference. In a character and symbol inspection device for inspecting by comparing with a reference image, an inspection image and a reference image which are captured multi-valued images are converted into a binary image, and an image area of an inspection target from the reference image converted into the binary image Image processing means for creating a plurality of different template types as a comparison reference, and the inspection image and the template are overlapped to extract a difference area, and a pass / fail judgment is made by comparing with a determination threshold value. And a determining means.

It becomes possible to identify and inspect defects such as bleeding, thinning, or chipping by comparison and determination using a template, and the inspection accuracy can be significantly improved compared to the conventional case. Further, the template is used to make a pass / fail determination in consideration of the line width of the inspection character, and a defect determination closer to a human determination standard can be made. 2 more inspection images
High-speed processing can be performed by converting the image into a value image, extracting the difference between the inspection image and the template, and comparing and judging with the judgment threshold value.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below in detail with reference to the drawings.

In FIG. 1, reference numeral 1 indicates a character / symbol inspection apparatus as a whole, and an image processing board 2 for image processing.
And a sequencer 3 for controlling a base machine (not shown) that conveys a semiconductor device on which characters to be inspected are printed.
I / O for sending and receiving image input signals and judgment signals to and from
The board 4 is connected to the CPU board 5. An inspection character imaged by the CCD camera 6 is captured in the image processing board 2, and an image can be displayed by the monitor 7. The inspection parameters can be set by operating the mouse 8 connected to the I / O board 4 while referring to the monitor 7.

On the other hand, a hard disk 9 and a floppy disk 10 are connected to the CPU board 5 as storage devices. The hard disk 9 stores a system control program for controlling the entire apparatus, a character inspection program used for character inspection, a menu program of the character inspection program, and a master image file in which a master image of the inspection character is filed. Floppy Pie Disk 10
The backup file of each file is stored in. Prior to the execution of the character inspection program, the character / symbol inspection device 1 registers, in the master image file of the hard disk 9, a character serving as a reference for comparison of characters to be inspected by the CCD camera 6 as a master character.

The actual character / symbol check is executed according to the character / symbol check sequence shown in FIG. That is, when the inspection system is started up, the character / symbol inspection sequence starts with the CPU of the CPU board 5 in step SP1.
By control, a master image that is a reference image of the inspection character is created from the character image captured by the CCD camera 6. Alternatively, when the master image is already registered, the master image is read from the hard disk 9 or the floppy disk 10 under the CPU control of the CPU board 5 and written in the image memory (not shown) of the image processing board 2.

In the next step SP2, the CPU of the image processing board 2 is controlled by the CPU to create various templates as a reference for image comparison in which the image area of the inspection character is modeled from the master image. Next, in step SP3, the inspection parameter for the character symbol inspection is set by operating the mouse 8 on the menu program read from the hard disk 9. That is, the inspection determination level, the number of re-inspections, etc. are set as the inspection parameters, and the inspection by the character / symbol inspection device 1 is made possible.

At the subsequent step SP4, the CPU
When the image capture signal from the sequencer 3 is received via the / O board 4, the character / symbol inspection is started. CPU
Captures the inspection image captured by the CCD camera 6 in step SP5 into the image memory of the image processing board 2,
At step SP6, the inspection character is cut out from the inspection image. Next, at step SP7, 1 is set on the image processing board 2.
The master character and the inspection character are aligned character by character, and the character symbol inspection using the template is executed in the subsequent step SP8.

At step SP9, if the inspection character is a non-defective product based on the comparison judgment result of the CPU master character and the inspection character based on the inspection program, the operation proceeds to step SP10. When the inspection of all the characters in the inspection target character string is completed, the process proceeds to the following step SP11, after which a determination signal indicating a non-defective product is sent out, and then the process returns to step SP4 to capture the image of the next inspection character string, and thereafter, step SP5. -SP6-
The inspection sequence of SP7-SP8-SP9 is executed.
If the inspection of all the characters in the inspection character string is not completed in step SP10, the process returns to step SP7 to execute the subsequent character symbol inspection sequence from the alignment of the next inspection character in the inspection image.

If it is determined in step SP9 that the inspection character is defective according to the determination comparison of the CPU, the process proceeds to step SP12. If the re-inspection is not designated here, the process proceeds to step SP11 to immediately send a determination signal indicating a defect. On the other hand, if the reinspection is designated, the process moves to step SP13 to judge whether or not the reinspection has been performed up to the designated number of times. If the reinspection has not been completed the specified number of times, the CPU control proceeds to step SP1.
After shifting to step 4, the start position of the alignment is changed, and then the process proceeds to step SP7 to execute the character symbol inspection sequence of steps SP8 to SP9 again from the alignment of the inspection character.

If the re-inspection has been completed up to the specified number of times at step SP13, the CPU control is made at step SP11.
Then, the judgment signal indicating the defect is immediately sent out, and then the steps SP4-SP5-SP6-SP7-SP8-SP9 are executed.
Perform the character-sign check sequence of.

Registration of the master character and creation of the template are executed in accordance with the master character registration and template creation procedure shown in FIG. First, in step SP20, the character string in the master image is surrounded by a window with the mouse 8. As a result, in step SP21, the characters in the window are cut out by the horizontal and vertical profiles by the master image registration program in the character inspection program.

If there are other characters to be registered, the process returns from step SP22 to step SP20 and the same processing is executed. When the character registration is completed, the process proceeds to step SP23. In step SP23, as shown in FIG. 4, the master image MA is converted into a binary image to create the standard template 20. In the subsequent step SP24, the expansion template 21 for bleeding inspection is created by performing expansion processing for thickening the contour of the character by one pixel on the standard template 20. In the next step SP25, the contraction template 22 for character thinning inspection is created by subjecting the standard template 20 to contraction processing for narrowing the contour of the character pixel by pixel. Each template created in this way is stored in the image memory of the image processing board 2. Since the number of expansion and contraction processes can be selectively set by the menu program, the number of expansion and contraction processes can be changed to perform an inspection assuming the limit of defective bleeding and fineness.

Next, the character / symbol inspection by the template will be described according to the inspection sequence of the inspection unit shown in FIG. In the inspection sequence, first, in step SP30, the inspection character is binarized, and then bleeding determination in step SP31, fineness determination in step SP32, and step SP33.
The process for determining the lack of is executed. Hereinafter, the process of each step will be described with respect to each of the inspection characters 30, 40, and 50 having bleeding, fineness, or lacking by using the inspection character "I" as an example.

First, the bleeding determination in step SP31 will be described according to the bleeding determination sequence shown in FIG. In the bleeding determination, the difference between the dilated template 21 and the inspection character 30 binarized in step SP41 is calculated. As a result, the bleeding portion 31 of the inspection character 30 which is outside the expansion template 21 is extracted. The area of the bleeding portion 31 extracted in the next step SP42 is calculated from the number of pixels. Then, in step SP43, the CPU control is performed to compare and determine whether the area of the extracted bleeding portion 31 is equal to or more than the threshold N for bleeding determination or less.

Next, the thinness determination of step SP32 will be described in accordance with the thinness determination sequence shown in FIG. In the thinness determination, first, the difference between the contraction template 22 and the inspection character 40 that has been binarized in step SP51 is calculated. As a result, the thin portion 41 of the inspection character 40 that is inside the contraction template 22 is extracted. At the next step SP52, the area of the extracted thin portion 41 is calculated from the number of pixels. Subsequently, in step SP53, the CPU controls to compare and determine whether the area of the extracted thin portion 41 is equal to or more than the threshold N for thin determination or less. Here, the thinness determination threshold value H is set as a ratio of the area of the thin portion 41 to the area of the entire master character.

Lastly, the chipping determination in step SP33 will be described according to the chipping determination sequence shown in FIG. In the determination of the missing portion, first, the difference between the standard template 20 and the inspection character 50 binarized in step SP61 is calculated. As a result, the defective portion 51 of the inspection character 50 inside the standard template 20 is extracted. Next, the image of the defective portion 51 extracted in step SP62 is subjected to shrinkage processing to generate a shrinkage-processed image 52 in which a portion having no thickness is deleted. The contraction processing image 52 is subjected to expansion processing at step SP63 to extract only the missing portion 53.

In the next step SP64, the area of the extracted missing portion 53 is calculated from the number of pixels. Then step SP
At 65, it is determined by comparison whether the area of the chipped portion 53 extracted by the CPU control is equal to or larger than the threshold value K for chipping determination or smaller. At this time, the chipping determination threshold value K is set by the ratio of the chipping area to the area of the square, with the area of the square having the line width of the master character as one side. This is because it is convenient to compare and determine a chipping defect that occurs as a “lump” by using the area of a square having the line width of the inspection character as one side to determine whether the quality is good or bad. The line width of the inspection character used here is that which has been calculated in advance using the standard template 20 when the master character is registered (SP1).

Here, the threshold value N, the threshold value H, and the threshold value K for the bleeding determination, the fineness determination, and the chipping determination can be set by a menu program. In this way, the character / symbol inspection can be processed at high speed by the simple comparison and determination in which the template created based on the master character and the inspection character are binarized to obtain the difference. Further, by using three types of templates, the standard template 20, the expansion template 21 and the contraction template 22, to identify and inspect defects such as bleeding, thinning or chipping, it is possible to significantly improve the inspection accuracy. Further, by using the template and inspecting while considering the line width of the inspection character, it is possible to make a defect determination close to a human criterion.

In the character / symbol inspection, the master image is first created and written in the image memory of the image processing board 2 in step SP1 of the character / symbol inspection sequence. If the master image is already registered at this time, the master image is read out through the hard disk 9 or the floppy disk 10 under the control of the CPU.

Next, in step SP2, the standard template 20, the expansion template 21 and the contraction template 22 are created from the master image by the CPU control. Here, the standard template 20 is created by binarizing the master image. Further, an expansion template 21 obtained by performing expansion processing on the standard template 20 is created as a template for bleeding inspection, and a shrinkage template 22 obtained by performing shrinkage processing on the standard template 20 is created as a template for fineness inspection.

In the character / symbol inspection sequence, after the inspection parameters are set (SP3), when the image capture signal is received (SP4), the inspection image is captured (SP).
5). Subsequently, the inspection character is cut out (SP6), the characters are aligned (SP7), and then the inspection using the template is executed in step SP8. In step SP8,
The inspection character input as a multi-valued image is converted into a binary image, and bleeding determination, fineness determination, and chipping determination are executed. First, the bleeding determination is performed by the expansion template 21 and the inspection character
The bleeding portion 31 extending outside the expansion template 21 is extracted from the difference from 0 and compared with the bleeding determination threshold value N, and the quality is determined.

Next, in the thinness determination, a pass / fail portion 41 that is inside the shrinkable template 22 is extracted from the difference between the shrinkage template 22 and the inspection character 40, and a pass / fail judgment is made by comparison with a thinning determination threshold value H. Further, the determination of chipping is performed based on the difference between the standard template 20 and the inspection character 50.
After 1 is extracted, shrinkage processing is performed to generate a shrinkage-processed image 52 of only the defective portion in which the portion having no thickness is deleted. By performing expansion processing on the contraction processing image 52,
The area of the notch 53 is calculated by extracting only the notch 53,
The quality of the inspection character 50 is determined by comparing it with the threshold value K for the defect determination based on the line width of the inspection character.

According to the above configuration, the standard template 2
0, an expansion template 21 and a contraction template 22 are used to identify and inspect defects such as bleeding, thinning or chipping by inspection using three types of templates, which is a significantly improved inspection compared to the conventional one. The accuracy can be realized and the reliability of the inspection can be improved. Further, according to the present invention, the template can be used to set a criterion in consideration of the line width of the inspection character, and thus a defect determination close to the human criterion can be made.

Furthermore, according to the present invention, image processing can be performed at high speed by converting the image of the inspection character into a binary image and processing it. Further, by the matching process using the template, the defective portion of the inspection character can be easily extracted from the difference area. Further, by using the area of the difference region as a parameter for comparison and determination, the determination standard can be easily set, and the determination threshold value can be easily set.
As a result, the algorithm of the conventional inspection program can be shortened by several tens of percent or more, and the speed of character symbol inspection can be increased.

In the above embodiment, the case where the shape inspection target is a character printed on the surface of the semiconductor device has been described, but the present invention is not limited to this, and a character or a number to be displayed as an image is displayed. If the shape can be discriminated by binary image processing such as symbols and figures, the same effect as the above case can be obtained.

[0034]

As described above, according to the present invention, it becomes possible to identify and inspect defective defects such as bleeding, fineness or chipping by using a plurality of different template types, which is far superior to the prior art. The inspection accuracy can be improved. Further, by using the template to determine the defect in consideration of the line width of the inspection character, it is possible to make the defect determination closer to the human criterion. Further, a character and symbol inspection apparatus and method capable of speeding up the character and symbol inspection by a simple inspection of converting a test image into a binary image and then extracting a difference area by using a template to detect a defective portion. realizable.

[Brief description of drawings]

FIG. 1 is a schematic diagram showing an overall configuration of a character / symbol inspection apparatus according to an embodiment of the present invention.

FIG. 2 is a flow chart showing a character / symbol check sequence.

FIG. 3 is a flowchart showing a procedure for registering a master character and creating a template.

FIG. 4 is a schematic diagram for explaining the creation of a template.

FIG. 5 is a flow chart showing an inspection sequence of an inspection unit.

FIG. 6 is a flow chart showing a bleeding determination sequence.

FIG. 7 is a flow chart showing a fineness determination sequence.

FIG. 8 is a flow chart showing a defect determination sequence.

[Explanation of symbols]

1 ... Character / symbol inspection device, 2 ... Image processing board, 3 ...
... Sequencer, 4 ... I / O board, 5 ... CPU board, 6 ... CCD camera, 7 ... Monitor, 8 ... Mouse, 9 ... Hard disk, 10 ... Flotpie disk.

Claims (4)

[Claims] 1. A character / symbol inspection device for inspecting by comparing an inspection image obtained by picking up an image of an inspection object formed of printed or displayed characters or symbols and performing image processing with a reference image as a comparison reference, An image conversion unit that converts the inspection image and the reference image, which are captured multi-valued images, into a binary image, and a comparison reference in which the image area of the inspection target is modeled from the reference image converted into the binary image. An image processing unit having a template creating unit for creating a plurality of different template types, and a difference region is extracted by superimposing the inspection image and the template on each other, and is compared with a determination threshold value to make a pass / fail judgment comparison. A character and symbol inspection device comprising: a determination means. 2. The template includes an expansion template for extracting a blur of the inspection target, a contraction template for extracting a fineness of the inspection target, and a standard template for extracting a chip of the inspection target. The character symbol inspection apparatus according to claim 1, further comprising: 3. A character and symbol inspection method for inspecting by comparing an inspection image obtained by picking up an image of an inspection target made up of printed or displayed characters and symbols and performing image processing with a reference image serving as a comparison reference, An image conversion step for converting the inspection image and the reference image, which are captured multi-valued images, into a binary image, and a comparison reference in which the image area of the inspection target is modeled from the reference image converted into the binary image. It comprises a template creation step for creating a plurality of different template types, and a comparison judgment step for judging the quality by superposing the inspection image and the template to extract a difference area and comparing the difference area with a judgment threshold value. Character and symbol inspection method characterized by the following. 4. The template creation step includes a blur extraction step for calculating a difference between an expansion template and the inspection image, and extracting a blur of the inspection target outside the expansion template as a difference area, a contraction template and the above The difference between the standard image and the inspection image is obtained by taking the difference between the standard image and the standard image and the fineness extraction step that extracts the fineness of the target to be inspected inside the contraction template as a difference area. A defect extraction step of extracting a defective portion, performing shrinkage processing on the defective portion, deleting a portion having no thickness, and performing expansion processing to extract the defective portion as a difference area. The character and symbol inspection method described in 3.