JPS6228886A - Method for discriminating object - Google Patents

Abstract

PURPOSE:To make it possible to discriminate an object even if a phase difference between patterns to be discriminated is not enough large by extracting a point to be a feature on a line graphic at first and then dividing said line graphic into plural segments by using the extracted point as a starting point. CONSTITUTION:A processor 17 traces the outline of an object, graphs the outline and searches a feature point on the pattern. Segmenting processing for dividing the line graphic into plural segments by using the feature point as a starting point is executed and the segment information including data inherent in each segment and inter-segment connection/array data is extracted in each segment. In a dictionary registration mode, said processing is executed in each reference object, and in a test mode, similar processing is executed for an unknown object. Respective extracted information are stored in a prescribed memory. Respective normalized data of the reference objects and unknown object stored in the memory are successively collated along the line graphic while matching their starting point each other and the reference object having the highest similarity is decided as the reference object for the unknown object.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an identification method for identifying (inspecting) an object by converting it into a line diagram using an image processing device.

[Conventional technology]

Recently, the need for object identification has increased in various industrial fields. For example, this is the case when the component parts of a product are identified and non-linked in order to automatically assemble the product. Conventionally, as such an object identification method, the so-called 5RI (Stanford Re5ea) has been used.
A matching method and a matching method are known.

The matching method is a method of extracting all data related to the entire or part of an identification object pattern and comparing it with a standard pattern for identification.

However, this method has b) It is difficult to apply this method unless the posture of the object is good.

(Example) It is impossible to identify objects unless they are separated from each other.

Not only are there problems and constraints, but the amount of data required to handle the object pattern itself is enormous, and the larger the changes in the posture of the object (the more complicated and difficult the position normalization and matching process become). .

The SRI method is as follows.

FIG. 7 is a diagram for explaining the SRI method.

This involves first encoding a binary image of the object using run lengths (segments), and then performing connectivity analysis on the run length data to extract various feature amounts. Then, useful parameters are calculated based on the basic feature values and their combinations.There are many basic feature values, but for example, the total area S and the circumference of the object 10 as shown in FIG. L, the number of holes H in the object 10, the sum of the areas of each hole H, the center of gravity Pc (Xe, yc) of the object 10, the maximum value RmaX% of the radius from the center of gravity to the contour, and the minimum value Rmin
- Projection maximum value XmaX and minimum value Xm1n onto the X axis,
There are a maximum value YmaX and a minimum value Ymin projected onto the Y axis. In addition, as the combination parameters of basic characteristics, for example, Pe (Mi4°S/L"), Rr (=Rmax
Rmin), Xd(EEXmax Xm1n)-y
d (=ymaX Ymin), etc. In addition,
The identification process is performed by showing each object in advance.
After registering the information by showing it to an image processing device), the parameters are compared and verified with the object to be identified.

[Problem 1 to be Solved by the Invention As described above, the SRI method is a method that focuses on the geometric features of an object, but it has the following problems.

a) Identification cannot be made unless the difference in shape between the patterns to be identified is large.

b) The objects to be identified cannot be identified unless they are separated from each other.

In other words, a) is a constraint on the type of objects that can be identified and the complexity of their shapes, and b) is a constraint on the arrangement and background of each object, which requires that the objects do not overlap or combine with each other. The condition is that no other patterns exist in the background.

Therefore, this invention solves the following problems: 1) Even if the shape difference between the patterns to be identified is too large, it can be dealt with.・2) Not only the objects to be identified are separated from each other, but also the cases where they overlap. Be able to deal with any situation.

3) Improve identification accuracy and shorten identification time (save shape features and compress data).

The purpose is to

[Failure to solve the problem]

An image processing device is provided that captures an object as an image, extracts various feature amounts of the object from the image information, and performs predetermined processing based on this.

[Effect]

The image processing device performs the following processing.

1) Trace the contour of the object and convert it into a line figure.

2) Search for characteristic points (feature points) on the line figure.

These characteristic points include breaking points and uneven points where the curvature changes rapidly.

3) Perform line segmentation processing to divide the line figure into a plurality of line segments using the feature point as a starting point, and extract line segment information including line segment-specific data and line segment connection/arrangement data for each line segment. In this case, the line segment-specific data includes the length of the line segment, the coordinates of both end points of the line segment, etc., and the connection/arrangement data of line segments includes the angle formed by each line segment, the representative point of each line segment (
distance between end points, center point), etc.

4) In the dictionary registration mode, the above-described processing is performed for each standard object, and various information extracted as a result is stored in a predetermined memory.

5) In the inspection mode, the same process is performed for the unknown object, and various extracted information is stored in a predetermined memory.

6) The normalized data of the standard object and the unknown object stored in the memory are matched sequentially along the line shape and the above starting points are made to match each other, and the standard object with the greatest degree of similarity is obtained. is determined to be an unknown object.

〔Example〕

Figure 1 is a reference diagram for explaining the invention in detail;
FIG. 6 is a block diagram showing an image processing apparatus in which the present invention is implemented, FIG. 6 is a reference diagram showing an example of storing line segment information in memory obtained by the present invention, and FIG. 4 is a contour line consisting of a series of pixels. FIG. 5 is a reference diagram for explaining a direction code, and FIG. 6 is a reference diagram showing an example of storing contour information in a memory.

First, explanation will be given starting from FIG. In the figure, 10 is an object to be identified, 11 is an illuminator, 12 is an imaging device such as a television camera, 15 is a log/digital (A/D) converter, and 15 is a log/digital (A/D) converter.
14 is a preprocessing circuit, 15 is a feature extraction circuit, 16 is an image memory, and 17 is a processing device such as a microprocessor.

The optical image of the occupational object 10 illuminated by the illuminator 11 is converted into a video signal by the television camera 12. This video signal is converted into a digital quantity for each pixel by an A/D converter 15, and further processed by a preprocessing circuit 14.
Noise removal, smoothing, etc. are performed on the image, and the image is extracted as a digital image. The feature extraction circuit 15 extracts various feature quantities from this digital image and stores them in the image memory 16. The processing device 17 performs connectivity analysis by 4-connection or 8-connection of 4 or 8 pixels adjacent to each pixel of interest, and extracts each prop (Blob).

A prop represents an independent pattern and is separated from the background.

Next, the contour of each prop is extracted and contour tracing is performed. Information regarding this contour line is stored in the image memory 16.

Now, suppose that a part of the contour line is represented as a series of pixels as shown in FIG. It can be expressed by the direction code Do-D7. Here, the direction code represents the direction (vector) when tracing the contour line pixel by pixel.
There are eight directions as shown by DO to D7 in the figure. Note that the angle between each direction vector is 45°. Further, these contour data are stored in the image memory 16 as shown in FIG.

There are many possible ways to use this direction code; for example, it can be used to calculate the perimeter along a contour line. That is, the circumference lengths from point 21 to point 29 in Figure 4 are 0°, 90°, and 180°, assuming the direction code DO is 0°.
°, 270°, each has a length t of one pixel,
The combinations of direction codes of 45°, 135', 225°, and 315' can be considered as the length t of one pixel multiplied by 7, and can be calculated as the sum of each part. Therefore, the perimeter from point 21 to point P9 in Figure 4 is (sxl-+-5x
v'T)xz.

In this way, the image pattern of the object is converted into a line figure using contour lines. Description (representation) of the object using this contour line
allows for significant data compression while containing sufficient shape information and feature information of the object. In addition to the method of using a television camera as described above, methods of using a flying spot scanner or laser light may also be considered for tracing the contour line.

Next, line segmentation is performed to divide the contour line.

Although various methods of line segmentation can be considered, an example according to the present invention is shown in FIG.

First, the outline of the line figure is traced to find characteristic points (feature points) or characteristic small sections. Point P1 in Figure 1 (a)
~P4 is a feature point or a characteristic small section, Pl is a break point, and P2 to P4 are convex points.

In other words, if we examine the change in the curvature (reciprocal of the radius) during the process of tracing contour line 1, we find that the break point P1 is as shown in Figure 1 (b).
As shown in the figure, the curvature changes discontinuously from 0 to (1), and the peak point P5 among the convex points can be regarded as the point where the curvature changes continuously and significantly. can. Therefore, such a feature point or a characteristic small section is adopted as a starting point for the line segmentation process.

Note that in the case of the corner point P10, this can be used as is, but for convex points, either the peak point P3 or the end point P4 is selected as the starting point.

Next, the contour line 1 is divided into line segments L1eL2, . . . . What is considered at this time is the tangent line of each point.

For example, the outline (part) 1 shown in Figure (C)
The tangent line 2 at 25 points can be found as y-ys-a(x-xs), where its x and y coordinates are X5ey5. Here, a represents the inclination (gradient) of the tangent line 2, and is determined by an appropriate method. Then, point P6 (x6
．． Let y6) be the end point of one line segment. This P6 (x6.
The point y6) is the end point of one line segment L2, but it is also the starting point of the next line segment L3, and the following line segment L
5 + La... can be obtained. Incidentally, at a point such as the corner point P1 in FIG. 3C, where the above method is meaningless, the next point, for example, P5, can be determined by a convenient method. Another method of line differentiation is to calculate the amount of deviation d (t or θ0) between the contour line and the tangent line.
) and the peripheral length on the contour line. In this case as well, the point on the contour line that reaches the predetermined tolerance ratio is set as the end point of the line segment. In any case, by such line segmentation, data of one continuous contour line, that is, object data, can be significantly compressed.

The line figure obtained in this way is described using line segment information. In this case, line segment information refers to (a) line segment-specific information port) connection/arrangement information for each line segment. More specifically, a) is the label of the line segment, the length of the line segment, the coordinates of both end points of the line segment, etc.; The distance between representative points (end points, center points, etc.) of

This line segment information is also stored in a predetermined memory, an example of which is shown in FIG. Note that all of the data in A) 1) is not necessary; for example, the label in B) and the data other than the connection head in A) can be omitted by associating them with addresses in the memory. In addition, the coordinates of both end points in (a) are one set of information, and the line segment itself is momme information (unique information).
In addition, it also serves as information representing the connection relationship with adjacent line segments, and is effective as original data before normalizing the position of the target figure. If you use the angle θ (see Figure 1 (c)) between the mouth and the adjacent line segment, the amount of stored data will be reduced and processing will be easier, but this is because the angle θ depends on the orientation of the target figure. This is because the data is not normalized. This is the length of the line segment a) 11*
The same applies to t2...).

Next, the identification process will be explained.

Prior to job-specific processing of unknown objects (also referred to as inspection mode), an operation is first performed to learn and register all standard objects (also referred to as dictionary registration mode). At this time, each object is imaged separately from each other in a background with good contrast, each of these objects is converted into a line shape as described above, feature points are extracted, and line segmentation processing is performed using these points as a reference. By doing this, line segment information is extracted. Feature point information and line segment information of each standard object are registered in a predetermined memory.

In the inspection mode, the same process as above is performed on the unknown object, and its feature point information and line segment information are registered in a predetermined memory.

Each extraction information of the standard object and the inspection object is compared with each other along the contour line with priority given to the information on the feature points, and the standard object that gives the highest degree of similarity is determined to be the applicable object of the inspection object. do. Here, regarding line segment information regarding one line segment, if that of the standard object is expressed by f and that of the inspection object by f', then the degree of similarity S can be expressed, for example, as in the following equation.

S-Cf-1f-f'l )/f Therefore, if both features if and f' match (f'
-f), S=l. Similarly, if the similarity of the nth line segment is Sn, and wnf is the weighting coefficient (ratio to the whole) of the nth line segment, then the similarity St for all line segments (number of line segments; N) is generally as follows. can be converted into Note that if f'-f holds for all parts, the degree of classification of all line segments is also It I
It becomes It, and can be completely matched and disappear.

〔Effect of the invention〕

According to this invention, by extracting contour information of an object and treating the object as a line figure consisting of a plurality of line segments, it is possible to perform extensive data compression without sufficiently extracting the shape features of the object. , a point A is brought about in which the memory capacity can be reduced and the calculation time can be shortened. In addition, feature points (v points and uneven points) on line figures are extracted and used as the first criterion for line segmentation, so not only can stable line segment information be obtained at all times, but also Improving the efficiency of matching (reducing unnecessary matching) - Brings about possible benefits.

As a result, the present invention has the following advantages: a) The shape of the object is not simple, and there are many types of objects.

(Example) We want to identify not only separated objects but also overlapping objects (partial patterns).

It is possible to meet such needs. In other words, needs such as (a) and (2) above are met by various types of automation.
(FA * OA + LA)
・While these problems are gradually becoming stronger, conventional SRI methods and matching methods cannot deal with them:
Therefore, the present invention has been made to meet these needs.

[Brief explanation of drawings]

FIG. 1 is a reference diagram for explaining the present invention in detail, FIG. 2 is a block diagram showing the configuration of an image processing device in which the present invention is implemented, and FIG. 3 is a diagram showing line segment information obtained by the present invention. FIG. 4 is a reference diagram showing an example of a contour line made up of a series of pixels, FIG. 5 is a reference diagram for explaining direction codes, and FIG. 6 is a reference diagram showing contour information. A reference diagram showing an example of storage in a memory, and FIG. 7 is a reference diagram for explaining the SRI method. Code explanation 1... Contour line, 2... Tangent line, 1o...
... Identification target part, 11 ... Illumination BA device, 12
...Song TV camera, 13...A/D converter,
14... Front processing circuit, 15... Feature extraction circuit, 16... Image memory, 17... Processing device. Agent Patent Attorney Akio Namiki Agent Patent Attorney Matsuzaki Mizogata Figure 1 Figure 2 @ Figure 3 WE 4 Figures etc. 5 Figure z Figure 6

Claims (1)

[Claims] A method of identifying the object by capturing an image of the object, processing the image, converting the object into a line figure, and performing predetermined processing based on the line figure, the method being a method of identifying the object, the feature of which is on the line figure. A process of first extracting , dividing the line figure into multiple line segments using this point as a starting point, and extracting line segment information for each line segment, including data specific to that line segment and connection/sequence data between line segments. After performing the above for each standard object, the same process is performed for the unknown object, and the resulting line segment information for both is made to coincide with each other, and then sequentially processed along the line figure. 1. An object identification method characterized in that a predetermined standard object from which the greatest degree of similarity is obtained is determined as a corresponding object of an unknown object.