JPH01229384A - Pattern matching device - Google Patents

Abstract

PURPOSE:To speed up pattern matching when the pattern matching of picture data capable of containing dislocation due to rotation is performed by evaluating the dislocation of a position without performing such processing as to rotate the picture data. CONSTITUTION:When a keyboard 1 is installed at a correct position below a video camera 3, image pickup is started, and a video signal is stored in a video RAM 13 while being developed. Subsequently, the picture data of a specified key is read out. Afterwards, corresponding sample data stored previously in a ROM 11 is read out. Both data are compared, and the number of discordance is counted. Afterwards, correcting processing using correction data is performed. Subsequently, the total number of the discordance is calculated, and whether the number is below a predetermined reference value or nor is decided. Namely, the dislocation of the picture data from the sample pattern is evaluated according to the total number of times of discordance after correction corresponding to a distance from a center of rotation in place of the evaluation by the number of times of simple discordance.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention performs pattern matching by evaluating the positional deviation between input image data and sample pattern data prepared in advance. Regarding equipment.

[Prior Art] Conventionally, in this type of pattern matching device, in order to suitably perform pattern matching when an object to be image processed is shifted from a reference position, positional shift correction is performed. I am doing it. In particular, in the case of misalignment due to rotation, even a slight rotation can significantly degrade pattern matching, so a device that evaluates matching by rotating and correcting image data input by imaging, etc. little by little, or performing coordinate transformation. is common.

[Problems to be Solved by the Invention]However, in such a pattern matching device,
The processing load of rotating and correcting input image data or sample patterns and converting coordinates is large, which has been an obstacle to realizing high-speed pattern matching processing. As a result, there is a problem in that product inspection using pattern matching takes a long time.

In particular, when the rotation center of the object is known and the rotation range is limited, in other words, if it is not matched within that range, it can be judged as defective, then the rotational deviation can be corrected. There is a problem in that performing large-scale processing such as rotation of image data in order to do so creates an excessive burden.

The present invention aims to improve the evaluation of pattern matching and solve the above problems.

The structure of the present invention that achieves the above object will be explained below.

[Means for Solving the Problems] A pattern matching device of the present invention performs pattern matching by evaluating a positional shift between input image data and data of a sample pattern prepared in advance. The present invention is characterized in that it includes a correction means that evaluates the deviation to be smaller as the distance from the predetermined center of rotation in the sample pattern increases.

[Function] The pattern matching device of the present invention having the above configuration is as follows:
For example, a positional shift between image data captured and input by an imaging means such as a camera and data of a sample pattern prepared in advance is evaluated, and pattern matching is performed. In this evaluation, the pattern matching device of the present invention uses the correction means to evaluate the deviation to be smaller as the distance from the predetermined rotation center in the sample pattern increases.

Therefore, the pattern matching device of the present invention performs pattern matching while relatively minimizing the evaluation of large deviations in the image periphery caused by slight rotation, and does not require processing such as rotation of image data.

[Example] In order to further clarify the structure and operation of the present invention described above, preferred embodiments of the pattern matching device of the present invention will be described below. The pattern matching device of the embodiment is used as a part of a keyboard inspection device.

As shown in FIG. 1, the keyboard inspection device includes a video camera 3 using a CCD that takes an image of the keyboard 1, and an electronic control system that processes the video signal from the video camera 3 to determine whether the keyboard 1 is good or bad. It is composed of a device 5. The electronic control device 5 includes a well-known CPU10. ROMII
, RAM12. It is configured as a so-called arithmetic and logic operation circuit equipped with a video RAM 13, etc., and is connected via a bus 14 to a video signal input board 15, a CRTC 17 that controls a display 16, an output board 20 that drives an alarm device 1, a printer 19, etc. and are interconnected.

Inside the ROM II, in addition to a program for executing control, data of sample patterns to be used for pattern matching, which will be described later, and correction parameters for evaluation are stored in advance.

Of course, the sample data etc. may be configured to be supplied by media such as a magnetic storage medium or an IC card.

A video camera 3 is connected to the video signal input port 15, and the video signal output from the video camera 3 is converted into a digital signal in real time, further converted into a digital signal, and stored in the video RAM 13. Therefore, by accessing a predetermined area of the video RAM 13, the CPU 10
Image data corresponding to a predetermined area of the keyboard 1 can be read out.

Next, the pattern matching process executed by the electronic device 5 will be described. FIG. 2 is a flowchart showing a control routine executed by CPU10. This control routine is started when the keyboard 1 is placed in the correct position under the video camera 3 by a conveyor 30 or the like.
Start imaging. As a result, the video signal output from the video camera 3 is input via the video signal input port, expanded and stored in the video RAM 13 (step S10'O). Next, a specific key, for example "
The process of reading out the image data of the "A" key is performed (step 5ilo). Since the position of the keyboard 1 with respect to the video camera 3 is predetermined, by reading data in a specific area, image data including characters engraved on the key top of a specific key can be obtained as shown in FIG. 10
It can be read out as a dot matrix with a size of approximately 10x10.

Thereafter, a process of reading sample data stored in advance in the ROMII is executed (step 5120). An example of a sample pattern corresponding to key rAJ is shown in FIG.

After reading out both data in this way, a process is executed to compare the actually captured image data and the sample pattern and count the number of discrepancies eq between the two (step 51).
30), the number of mismatches is then counted in the areas AI to A5@. As shown in FIG. 5, areas AI to A5 are areas around the rotation center RC of the key (see FIG. 3).
These are areas set concentrically, and each has a size of [4, 12, 20, 28, 36] in number of dots from the inside (AI side). For example, the image 1ff- shown in FIGS. 3 and 4 is
'? As shown in Figure 5 of this post, in order from territory @AI, [0, 1° 4, 8
．． 3].

After counting the number of mismatches eq for each region in this way, a process is performed to correct the number of mismatches eq using correction data (correction parameters shown in FIG. 5). That is, the number of mismatches eq for each region is divided by the correction parameter to obtain corrected mismatch data. This correction process corrects deviations from the sample pattern due to image data rotation.
In view of the fact that the distance from the center of rotation RC becomes larger, this is equivalent to weighting to correct this.

Following the above-described correction process, each area AI to A5=i
The total number of discrepancies Neq is calculated by adding the discrepancy data of (Step 5150), and it is determined whether the total number of discrepancies Neq is less than or equal to a predetermined standard (iNrf) (Step 9160). Total negative WtNciw% false hori, image arc 9 after correction according to distance from rotation center instead of evaluating deviation by number
The deviation from the f-shaped pattern is evaluated.

If the total number of discrepancies Neq exceeds the standard (iNrf-t-), it is assumed that an incorrect key top is attached to the specific key of the keyboard 1 imaged by the video camera 3, and the image of that key top is sent to the CRTC 17. The process executes processing such as displaying the alarm on the display 16 via the output boat 20 and outputting the alarm by driving the alarm device 18 via the output boat 20 (step 5170).After the alarm is output, or when the total number of discrepancies Neq is If it is less than Nrf (step 51bO), the VC asks for the key of love and judges whether or not the horizontal placement has been completed (step 5190).

If all keys have not been checked, proceed to step 1.
Steps 5110 to 180) of returning to step 10 and repeating the above-mentioned process for other keys. When the inspection of all keys is completed, the printer 19 is driven via the output boat 20 to issue an inspection certificate. The process is executed (step 5190). Of course, during the inspection of all keys, if an incorrect key top is attached, etc., that fact will be output. Thereafter, the process exits to "END" to end this control routine.

According to the present embodiment described above, there is no need to perform processing such as rotation on the captured image data to deal with deviations caused by the manufacturing-permitted rotation of the key tops of the keyboard 1, and pattern mapping can be performed at extremely high speed. You can perform pinching. Therefore, the overhead of image processing on the CPU 10 can also be suppressed. As a result, the keyboard 1 can be inspected at high speed, reducing the time required for the inspection process. In the above-mentioned actual precedent, five areas AI to A5 are set concentrically with respect to the center of rotation RC, but it is not necessary to limit the time to concentrically. For example, As shown in FIG. 6, it is also effective to prepare several area patterns according to the shape of the sample pattern, etc. in order to improve the recognition rate. Furthermore, the number of areas may be increased or decreased.

Although the embodiments of the present invention have been described above, the present invention is not equally limited to these embodiments, and it goes without saying that the present invention can be implemented in various forms without departing from the gist of the present invention. As described in detail above, according to the pattern matching device of the present invention, when performing pattern matching of image data that may include deviation due to rotation, evaluation of the positional deviation is performed based on the image data, etc. This has an extremely excellent effect in that it can be executed without performing processing such as rotating the image.

Therefore, pattern matching can be performed at high speed, and the processing load can be reduced. As a result, if the pattern matching device of the present invention is used, inspections using pattern matching can be carried out in an extremely short time without reducing accuracy.

[Brief explanation of the drawing]

, FIG. 1 is a schematic configuration diagram of a keyboard inspection device incorporating a pattern matching device as an embodiment of the present invention, FIG. 2 is a flowchart showing a pattern matching control routine executed by the electronic control device 5, and FIG. FIG. 3 is an explanatory diagram showing an example of the image data of a specific key top of the captured keyboard 1, FIG. 4 is an explanatory diagram showing an example of sample data, and FIG. 5 is an explanatory diagram showing an example of the image data and the sample pattern. FIG. 6 is an explanatory diagram illustrating the method, and FIG. 6 is an explanatory diagram showing an example of another area pattern. 1...Keyboard 3...Video camera 5...Electronic control unit 11...ROM 13...Video RAM Agent Patent attorney Tsutomu Sadatsu (and 1 other person) Figure 3 Figure 4 Section 5

Claims (1)

[Claims] 1. In an apparatus that evaluates a positional deviation between input image data and data of a sample pattern prepared in advance and performs pattern matching, the further the sample pattern is separated from a predetermined center of rotation, A pattern matching device characterized by comprising a correction means for evaluating the deviation to be small.