JPH03282974A - Method for inspecting character - Google Patents

Abstract

PURPOSE:To quickly inspect whether characters annexed to a body to be inspected are correct or not by allowing respective image processors to scan respective ranges of divided image memories, count up the number of picture elements indicating black levels and execute the pattern matching of respective characters. CONSTITUTION:A part on which the certificate number of a certificate 12 is printed out is photographed by a TV camera 11 and its image is stored in image memories 13 to 17. A transputer(TP) 26 in a deciding device 20 outputs a command for executing primary segmentation(SG) to TPs 21 to 25, the sort of the command signal is discriminated and respective TPs 21 to 25 execute the primary SG by dividing each image into two parts. The TP 21 scans the image memory 13, counts up the number of black picture elements and sends the counted value to the TP 26. The TPs 22 to 25 also execute similar operation for the memories 14 to 17. When a boundary is detected, an image in the boundary is divided into five parts and the secondary SG of respective images is executed by the TPs 21 to 25. The TP 26 finds out the boundary from the sent count number and coordinate data and compares the boundary with the data stored in the memories to execute pattern matching.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a character inspection method for inspecting whether or not characters attached to an object to be inspected are correct.

(Prior Art) Conventionally, inspection devices have been used to inspect whether numbers have been printed correctly on printed matter (inspected objects) such as securities or lottery tickets.
The one shown in FIG. 6 is known.

In FIG. 6, 1 is a television camera that takes an image of printed matter 2 on which a number is printed, 3 is an image memory that stores the image data taken by the television camera, and 4 is an image of the number stored in the image memory (not shown). A processing device including a microcomputer or the like recognizes the number printed on the printed material based on image data and determines whether the correct number has been printed, and 5 is a display device that displays the result determined by the computer.

Now, when the television camera 1 captures an image of the numerical part that is the number printed on the printed matter 2, and this captured image G (see FIG. 7) is stored in the image memory 3, it is first processed by the processing device. 4 performs partial segmentation based on the image data stored in the image memory 3 to determine upper and lower boundary lines on which numbers are printed.

That is, the processing device 4 processes the image G as shown in FIG.
The image memory 3 is scanned in the horizontal direction indicated by the arrow El from top to bottom, and the number of black pixels is counted for each scan, and a number is printed from the counted number at the upper limit boundary. Find line A. Next, the image memory 3 is scanned so that the image G is sequentially scanned from the bottom in the horizontal direction indicated by the arrow E2, and the lower limit boundary line B on which the numbers are printed is determined in the same manner as above.

Next, the processing device 4 performs secondary segmentation.

In this secondary segmentation, the range of boundaries A and B is scanned vertically in order from left to right, boundaries CI to C, D, I, and V are obtained for each number in the same manner as above, and these boundaries are The position range of each number is determined from A, B, C, and D.

Then, pattern matching is performed between the image data within the above position range and the image data showing the shape of numbers stored in a memory (not shown) prepared in advance, and the characters within each position range are recognized. To go. A number is determined from these recognized numbers, and this determined number is compared with the number printed on the previous printed matter. If the result of this comparison is, for example, consecutive numbers, the correct number is on that printed matter. If it is determined that the printed matter is printed, and the numbers are not consecutive, it is determined that the correct number is not printed on the printed material.

(Problem to be Solved by the Invention) However, in the above-mentioned inspection device, primary and secondary segmentation, pattern matching, etc. are performed by one control device, so the inspection device performs the It takes a considerable amount of time to determine whether a number is printed.

For this reason, it is not possible to inspect the securities printed and sent out from the printing device one by one as they are printed.
Therefore, conventionally, the apparatus is installed in a processing line with a slow production speed, and only a part of the manufactured securities, etc., is extracted and inspected.

For this reason, there is a problem in that even if inappropriate numbers are printed on securities, etc. that are not extracted, it is not known.

(Purpose of the Invention) Therefore, this invention was made in view of the above problems, and its purpose is to test whether the characters attached to the object to be inspected are correct or not in a short time. The purpose of this invention is to provide a character inspection method that can perform the following tasks.

(Means for Solving the Problems) In order to achieve the above object, the present invention images characters attached to an object to be inspected using an imaging device, stores this imaged image data in an image memory, and The image memory is scanned corresponding to the horizontal and vertical directions of the image, the number of pixels corresponding to the black level is counted, the position range of each character is determined from this counted number, and the position range of each character is A character inspection method that performs pattern matching of image data in the corresponding image memory to recognize each character in each position range, and determines from each recognized character whether or not the character attached to the object to be inspected is correct. A plurality of the image memories are provided, a plurality of image processing devices and a control device for controlling each of the image processing devices are provided corresponding to each of the image memories, and an imaging device is provided in the plurality of image memories. Each captured image data is stored, the captured image is divided into multiple parts, and each image memory is divided into multiple parts corresponding to these divided images.
Each range of the divided image memory is assigned to each image processing device, and each assigned image processing device scans the range of the divided image memory and counts the number of pixels indicating the black level. , the respective count numbers are sent to the control device, the position range of the characters added by the control device is determined from these count numbers, and pattern matching is performed by each image processing device from the image data in the image memory of each of the determined position ranges. The character data recognized through this pattern matching is sent to the control device, and the control device determines whether or not the characters attached to the object to be inspected are correct based on the sent character data. It is characterized by

(Function) Since the present invention has the above-mentioned configuration, image data captured by an imaging device is stored in a plurality of image memories, and the range of the divided image memories is scanned by each image processing device allocated to the image memory. The number of pixels indicating the black level is counted, each count is sent to the control device, the control device calculates the position range of the attached character from these counts, and stores the image memory in each position range. Each image processing device performs pattern matching from image data.

The character data recognized by this pattern matching is sent to the control device, and the control device determines from the sent character data whether or not the characters have been correctly attached to the object to be inspected.

(Example) Hereinafter, an example of a character inspection device that implements the method of the present invention will be described based on the drawings.

In FIG. 1, 11 is a television camera (imaging device) that takes an image of the part where a number is printed, such as a security (object to be inspected) 12, and 13 to 17 store image data taken by the television camera. 20 is a determination device that determines whether or not the number printed on the security certificate 12 is correct from the image data stored in the image memories 13 to 17;
30 is a display device that displays the result determined by the determination device 2o.

As shown in FIG.
.about.25, and a transpuiter (control device) 26 for controlling the transputers 21 to 25, and the transpuiters 21 to 26 are connected to each other as shown in the figure.

The transputers 21 to 25 have a memory (not shown) in which image data representing characters of numbers 0 to 9 are stored as a dictionary. Furthermore, the transputers 21 to 26 are capable of simultaneously communicating data and processing data.

Next, the operation of each transducer 21 to 26 is described in 3.4.
This will be explained with reference to the flowchart shown in the figure.

The flowchart shown in FIG. 3 shows the operation of the transputer 26, and the flowchart shown in FIG. 4 shows the operation of the transputers 21-25. Note that the step is indicated as S in the drawing.

Now, when the portion of the security 12 on which the number is printed is imaged by the television camera 11, this imaged image P (see FIG. 5) is stored in each of the image memories 13-17.

In step 1, it is determined whether the image data captured by the television camera has been stored in each image memory 13 to 17. If no, the process returns to step 1; if YES, the process proceeds to step 2. Then, the transputer 26 outputs a command for causing the transputers 21 to 24 to perform primary segmentation.

On the other hand, the transputers 21 to 25 operate according to the flow chart shown in FIG. Go back and if yes, proceed to step 21. In step 21, the type of command signal from the transputer 26 is determined. Since the transducer 26 is outputting the primary segmentation command, the process advances to step 22. In step 22, primary segmentation is performed by each transducer 21-24.

Here, the primary segmentation is to divide the image P shown in FIG. This will be done on the 24th.

That is, the image memory 13 is scanned by the transducer 21 so that the image Pa is sequentially scanned in the horizontal direction H1 from above, and the number of black pixels is counted for each horizontal scan. Transputer 2 to count the number
Send to 6. Similarly, the transducer 22 scans the image Pa sequentially in the horizontal direction H2 from the bottom.
The image memory 14 is scanned and the number of black pixels is counted each time the image memory 14 is scanned in the horizontal direction, and the count number for each scan is transmitted to the transducer 26.

Image pb is also processed in the same manner as above. That is,
The transducer 23 scans the image memory 15 so that the image pb is sequentially scanned in the horizontal direction H3 from above, and counts the number of black pixels for each horizontal scan. is transmitted to the transducer 26. Similarly, the image memory 16 is scanned by the transducer 24 so that the image pb is sequentially scanned in the horizontal direction H4 from the bottom, and the number of black pixels is counted for each horizontal scan. The count number for each time is transmitted to the transducer 26 (step 23).

In this way, the primary segmentation is performed by the four transputers 21 to 24, so the processing speed is four times faster than the conventional processing speed.

On the other hand, the transputer 26 is connected to each transpeater 21.
In step 3), the count numbers transmitted from 24 to 24 are sequentially received, and in step 4 it is sequentially determined from the count numbers whether the numbers are on the boundary line between the printed upper and lower limits. Then step 3 until you find the border
, 4 are repeated.

By the way, since the transputer can perform communication and processing at the same time, it can receive the count number and determine the boundary line at the same time, and the processing in step 3.4 becomes faster.

If 2 is found on the boundary line Kl shown in Figure 5, step 5
In step 5, a command signal for performing secondary segmentation is transmitted from the transpuiter 26 to each transpeater 21 to 25.

The secondary segmentation performed here is to divide the image within 2 along the boundary line Kl into 5 parts from the broken line position as shown in FIG.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

That is, the image memory 13 is scanned by the transducer 21 so as to sequentially scan the image p1 in the vertical direction from the left and right, the number of black pixels is counted for each vertical scan, and the scanning position is determined along with the counted number. The indicated coordinate data is transmitted to the transputer 26 (step 24
．． 25).

Here, the coordinate data is, for example, the image P shown in FIG.
This is position data when the origin is the X point.

Similarly, image memories 14 to 17 are scanned by transputers 22 to 25 so that images p2 to p5 are sequentially scanned vertically from left to right, and the number of black pixels is counted for each vertical scan, The count number and coordinate data indicating the scanning position are transmitted to the transducer 26 (step 24.25).

In this way, the secondary segmentation is performed by the five transputers 21 to 25, so the processing speed is five times faster than the conventional processing speed.

The transpuiter 26 is connected to each transpewter 21 to 25.
The count number and coordinate data transmitted from the computer are received, and the boundary line Ll-LIO in each of the images pi to p5 is determined from the received count number (step 6). The processing speed here is also fast because communication and boundary line determination are performed at the same time.

Then, in step 7, each transducer 21 to 25
It is determined whether or not data has been received. If no, the process returns to step 7; if yes, the process proceeds to step 8.

In step 8, the boundary lines L1 to L are determined based on the coordinate data.
Sort IOs in coordinate order. That is, the boundary lines L1 to LIO are arranged as shown in FIG. and,
A command is sent to each of the transputers 21 to 25 to cause them to recognize numbers within the boundaries L1 to LIO (step 9). In this embodiment, the number 1 shown in FIG.
A command is outputted to cause the transducer 22 to recognize the digit 2, and to cause the transducers 23 to 25 to recognize the digits 3 to 5, respectively.

Each of the transputers 21 to 25 performs pattern matching by comparing black pixels within the assigned boundaries with image data (dictionary) indicating the shape of numbers stored in a memory (not shown) in advance. (Step 26). Note that within the boundary line referred to here is the boundary line Kl surrounding the numbers shown in Figure 5.
, to 2. This is within the range of Ll-LIO.

Pattern matching is performed by finding the number N of matching pixels between the black pixels of the image data representing the shape of a number and the black pixels within the boundary line, and finding the degree of similarity from the following equation.

Similarity=, ρ1 to m−n> However, m is the number of pixels within the boundary line, and n is the number of pixels in the dictionary showing one character.

Then, the degree of similarity is calculated for each number 0 to 9 in the dictionary, and the number 0 to 9 with the largest similarity value is recognized as a matching number. from there to the transducer 26 (step 27).

In this way, since pattern matching is performed using five transducers, pattern matching can be performed five times faster than the conventional processing speed.

In step 10, each transducer 21-25 performs pattern matching and sends the recognized numerical data to the transducer 26. In step 11, it is determined whether or not the transputer 26 has received the numerical data sent from each of the transputers 21 to 25. If no, the process returns to step 10, and if YES, the process proceeds to step 12.

In step 12, a number is obtained from the received numerical data, and the obtained number is compared with the previously inspected number. If the number is continuous with the previous number, it is determined that the security has the correct number printed on it. If the number is not consecutive to the previous number, it is determined that the security is not printed with the correct number, and this fact is displayed on the display 30, and step 1
Return to

As shown in the second point, since a transputer is used, data communication and data processing can be performed at the same time, and furthermore,
Processing speed is 4 times faster for primary segmentation, 5 times faster for secondary segmentation, and 5 times faster for pattern matching. Therefore, if the determination time from when the security is imaged to when the number is determined is simply calculated, it will be 100 times longer than the conventional method.

Therefore, by installing the above-mentioned device on a printing line, it is possible to inspect the numbers of printed securities one by one.

In the above embodiment, the present invention has been described as being applied to an inspection device that inspects securities numbers, but the present invention is not limited thereto, and can of course be applied to, for example, an inspection device that inspects characters other than numbers.

(Effects of the Invention) According to the present invention, a captured image is divided into a plurality of parts, each image memory is divided into a plurality of parts corresponding to these divided images, and each range of the divided image memory is divided into a plurality of parts. Each image processing device scans the divided image memory range, counts the number of pixels that indicate the black level, and performs pattern matching for each character. Since this is a system designed to perform this test, it is possible to check whether the characters affixed to the object to be inspected are correct in a short period of time, and therefore, it is possible to check the printed ticket numbers one by one. It becomes possible.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment for carrying out the method of the present invention, FIG. 2 is a block diagram showing the configuration of the determination device, and FIGS. 3 and 4 show the operation of the transputer. FIG. 5 is an explanatory diagram showing the segmentation method; FIG. 6 is a block diagram showing the configuration of a conventional character inspection device; FIG. 7 is an explanatory diagram showing the conventional segmentation method. 11... Television camera (imaging device) 13-17... Image memory 21-25... Transputer (image processing device) 2
6...Transputer (control device) Fig. 0 Fig. 4 Condolence 5 Fig. 5

Claims (1)

[Claims] Characters attached to the object to be inspected are imaged by an imaging device, the imaged image data is stored in an image memory, and the image data is stored in the image memory in correspondence with the horizontal and vertical directions of the imaged image. scan and count the number of pixels corresponding to the black level,
Find the position range of each character from this count,
Pattern matching is performed on the image data in the image memory corresponding to each of the determined position ranges to recognize each character in each position range, and it is determined from each recognized character whether or not the character affixed to the object to be inspected is correct. A character inspection method comprising: providing a plurality of image memories; providing a plurality of image processing devices corresponding to each image memory; and a control device controlling each of the image processing devices; Each of the image data captured by the imaging device is stored in the image memory, the captured image is divided into a plurality of parts, and each image memory is divided into a plurality of parts corresponding to these divided images.
Each range of the divided image memory is assigned to each image processing device, and each assigned image processing device scans the range of the divided image memory and counts the number of pixels indicating the black level. The count numbers of are sent to the control device, the position range of the characters attached by the control device is determined from these count numbers, and pattern matching is performed in each image processing device from the image data of the image memory of each of the determined position ranges. Character data recognized through this pattern matching is sent to a control device, and the control device determines whether or not the characters attached to the object to be inspected are correct based on the sent character data. Character inspection method.