JPH0351027B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method for recognizing the continuity of a contour of an object to be detected.

A method of detecting a target object using an imaging device such as a television camera is to find the outline of the target object based on the areas where the brightness of the image data from the imaging device suddenly changes, and then detect the appropriate target object. There is something I tried to do.

The procedure for object recognition using the conventional contour extraction method described above will be explained with reference to Figures 1 a to d. First, the original image taken with a television camera (Figure 1 a) is subjected to differential processing along the scanning line. , one contour candidate point where the brightness suddenly changes is extracted (FIG. 1b). next,
Performing differential processing (difference processing) between the extracted contour candidate point and the surrounding 8 pixels adjacent to this contour candidate point,
The pixel with the maximum value among these eight differential values is regarded as the next contour candidate point following this contour candidate point, and by repeating this operation, continuous contour points (contour line candidates) are obtained (Fig. 1 c ), and when these contour points are closed (Fig. 1d), they are considered to be one object.

In addition, during the process of tracking the next contour candidate point mentioned above, if the obtained maximum value does not meet the predetermined threshold value, it is assumed that there are no consecutive points,
At that point, the tracking operation ends.

However, in such conventional contour line extraction methods, there are several factors that impede tracking of contour candidate points: (1) Blooming due to metallic luster (see Figure 2 a) (2) Overlapping objects (see Figure 2 b) (3) An unclear image due to rust, dirt, etc. on the surface of the object. (4) Image disturbance due to electrical noise. However, in the conventional technology, only a narrow range of 8 pixels around the contour candidate point was tracked, so when the image becomes discontinuous due to blooming, rust, dirt, or noise, the discontinuity of the contour point The problem is that there is a high possibility of detecting continuity. In addition, in the conventional technology, one contour candidate point is first found and contour recognition is performed by differential tracking using this contour candidate point as the starting point. When a branch occurs due to such reasons, there is a problem that there is a high possibility that the target object will be missed, and that the processing takes time.

The present invention was made in view of the above circumstances, and
This reduces the possibility of determining contour points as discontinuous due to various tracking obstruction factors, improves recognition when images overlap, reduces processing time, and creates contour lines that are more resistant to noise. The purpose of this paper is to provide a method for recognizing continuity.

According to this invention, image data indicating a contour candidate point of an object to be detected is estimated, and image data of several pixels including the contour candidate point and an image of several pixels outside the contour candidate point are estimated from the position of the image data. data, and image data of several pixels inside the contour candidate points, calculate the average value of the brightness of these three types of image data, and calculate the average value for each of the above types between adjacent contour candidate points. The above objective is achieved by calculating the respective difference values and recognizing that the adjacent contour candidate points are continuous when at least one of the three difference values thus calculated is within a predetermined range. That's what I do.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

FIG. 3 is a schematic configuration diagram showing an example of a device for carrying out the contour line continuity recognition method according to the present invention, and shows a case where the ITV camera 2 is photographing a circular object 1 as the object to be detected. .
The ITV camera 2 photographs the circular object 1 in a predetermined field of view and outputs a video composite signal containing brightness signals of the input image to the synchronization separation circuit 3 and the A/D converter 4. The synchronization separation circuit 3 separates a synchronization signal from the input video composite signal, specifies the address of a random access memory array (RAM array) 5 based on this synchronization signal, and the A/D converter 4 separates the synchronization signal from the input video composite signal. The bright/dark signal of the video composite signal to be
It is converted into gradation image data and written to the specified address position. In this way the RAM
The array 5 stores one screen worth of image data indicating the brightness of the original image shown in FIG. In addition,
By specifying the X and Y addresses of the RAM array 5, arbitrary image data can be extracted.

On the other hand, the memory 6 stores a main program etc. for implementing the method of the present invention, and the central processing unit (CPU) 7 performs image processing of the image data in the RAM array 5 based on the main program contents. Execute.

Next, the processing procedure of this CPU 7 is performed according to the flowchart shown in FIG.
This will be explained with reference to Figures a to c.

First, the current image data (6th
From figure a, feature points of the object to be detected are extracted. Here, since the object to be detected is a circular object, the center position P ₀ (X ₀ , Y ₀ ) of the circle is determined as its feature point (FIG. 6b). Detection of this feature point is performed using RAM, for example.
The image data of the array 5 is differentially processed in the One possible method is to calculate the center position of a circle passing through a point.

Next, contour candidate points are estimated from the feature point P ₀ and the radius of the circle. For convenience, each contour candidate point is
It is expressed as Pi (i=1 to n) (Figure 6c). Next, several pixels Ci including the contour candidate point Pi, and the contour candidate point Pi
Several pixels Oi outside of the contour candidate point Pi and several pixels Ii inside the contour candidate point Pi are extracted (Fig. 6 c, d, e). At the time of extraction, it is necessary to extract the three types of image data so that they are aligned approximately in the normal direction to the locus of the outline candidate point. Note that the normal direction is determined by the following formula based on the relative positional relationship ^between the feature point P ₀ (X ₀ , Y ₀ ) and the contour candidate point Pi (Xi, _Yi ) /X ₀ −Xi (1) It can be obtained by the following, and the several pixels are extracted according to this direction. Figure 7 a, b and c are three pixels Ci including the contour candidate point Pi, and the contour candidate point
Three pixels Oi outside Pi and three pixels Ii inside the outline candidate point Pi are shown.

Next, average values i, i, and i of the brightness of the three types of image data Ci, Oi, and Ii are determined, respectively. Then, when i is 1, i is set to 2, the above process is executed again, and the process moves to the next step.

In this step, the above average values i, i, i and _i-1 , _i-
The difference values S _C , S _O , and S _I of ₁ and _i-1 are calculated using the following formula, S _C = C _i −C _i-1 S _O = O _i −O _i-1 S _I = I _i −I _i-1 ...calculated by (2). Next, it is determined whether each of the difference values S _C , S _O , and S _I obtained in this way is within a preset tolerance range (from the lower limit T _L to the upper limit T _H ), as shown in Figure 8. If all three difference values are outside the above range, it is determined that the adjacent contour candidate points are discontinuous, and then i is set to 1 to check the continuity or discontinuity of the next adjacent contour candidate points. Increment by 0 and execute the above process again.
On the other hand, when any one of the three difference values is within the above range, it is determined that the adjacent contour candidate points are continuous. That is, in Fig. 8, there is a point between C and D where the difference value is outside the above range, but if any of the other difference values is within the above range, then the adjacent contour candidate points are continuous. It is determined that the When it is determined that adjacent contour candidate points are continuous, i is incremented by 1 to check continuity/discontinuity between the next adjacent contour candidate points, and the above process is executed again.

In this way, when continuity or discontinuity among all contour candidate points is confirmed, the above process ends. Then, the object is recognized based on the continuity determination result.

Note that the object to be detected is not limited to circular objects, but can also be used to determine whether or not the contour line is continuous regarding the contour line of objects of various shapes or a part of the contour line as shown in Figure 9. The method of the present invention can be applied to. However, it is assumed that the contour candidate points are estimated by some means and are known.

As explained above, according to the present invention, contour candidate points can be tracked in three different wide areas: several pixels including the contour candidate points, several pixels outside the contour candidate points, and several pixels inside the contour candidate points. If at least one of the above three types of judgment results satisfies the judgment criteria, the contour candidate points are recognized as continuous, so blooming due to metallic luster, rust on the object surface,
The possibility that the contour line is determined to be discontinuous due to dirt or the like is reduced, and a decrease in the object recognition rate due to the above-mentioned causes can be reduced.

Since contour candidate points are estimated in advance and the tracking range is limited, tracking can be accomplished in a short processing time even when a branch occurs due to overlapping objects, etc., and the recognition rate in this case can be improved. can be improved.

When comparing the difference values to confirm continuity, the difference between the average values of image data of several pixels is used, so it is resistant to image disturbances caused by electrical noise.

It has the following effects.

[Brief explanation of drawings]

Figures 1 a to d are diagrams used to explain the procedure of object recognition using the conventional contour extraction method, and Figures 2 a and b are diagrams showing examples of factors that inhibit conventional contour tracing, respectively. , FIG. 3 is a schematic configuration diagram showing an example of a device for implementing the contour line continuity recognition method according to the present invention, and FIG. 4 is a diagram showing the RAM of FIG. 3.
FIG. 5 is a flowchart showing an example of the processing procedure of the central processing unit in FIG. 3, and FIGS.
The diagram used to explain the flowchart in Figure
Figures 7 a to c are diagrams each showing an example of several pixels including a contour candidate point, several pixels outside the contour candidate point, and several pixels inside the contour candidate point, and Figure 8 is a diagram showing the brightness and darkness around the entire circumference of the contour line. FIG. 9 is a graph showing an example of the degree average value, and FIG. 9 is a diagram showing an example of other contour candidate points. 1... Circular object, 2... ITV camera, 3... Synchronization separation circuit, 4... A/D converter, 5... RAM array,
6...Memory, 7...Central processing unit (CPU).

Claims (1)

[Claims] 1. Image data indicating contour candidate points of the detection target object is estimated in advance from image data indicating the brightness of one screen where the detection target object exists, and the contour is estimated from the position of the image data. Image data of several pixels including the candidate point, image data of several pixels outside the contour candidate point, and image data of several pixels inside the contour candidate point are extracted, and the average brightness of these three types of image data is calculated. Each value is calculated, and the difference between the average values for each of the above types is calculated between adjacent contour candidate points.
A contour line continuity recognition method that recognizes that the adjacent contour candidate points are continuous when at least one of the two difference values is within a predetermined range. 2 Estimation of image data indicating the contour candidate points is as follows:
2. The contour continuity recognition method according to claim 1, wherein when the contour of the object to be detected is circular, the contour is estimated based on its center of curvature and radius.