JPH05196424A - Pattern recognition method - Google Patents

Abstract

PURPOSE:To enable a binary image and a shaded image to be coped with and an image which is not uniformly bright to be recognized positively by performing a primary differential calculation of a data which is projected in a specified direction at a processing region which is set according to the inclination of an object for judging the size of the object. CONSTITUTION:An edge point which is obtained initially is detected for each line and an inclined straight line 12 is obtained by performing scanning in the direction of an arrow A within an inclination detection window 10 of a shaped image which is set from the size of an object and a center position 9 of a nozzle. A lead detection window 13 which is matched to the inclination of the straight line 12 is set and a region is determined. A data which is obtained by projecting a brightness value in a direction which is vertical to the inclination of the straight line 12 within the window 13 is subjected to primary differential calculation. The primary differential value is scanned from the direction of the arrow A, thus obtaining a judgment point 16 (17) which is larger (smaller) than a threshold value 14(15) with a positive (negative differential value. All pairs where the width is smaller than a lead width are extracted out of points 16 and 17 which are greatly in proximity and a coordinate zero-cross point 18 which reaches 0 while changing from a positive to a negative value which is obtained between the points 16 and 17 indicates the center position of a lead 7.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern recognition method for detecting the lead position of a leaded component in a mounting machine or the like for mounting components on a circuit board.

[0002]

2. Description of the Related Art An example of a conventional pattern recognition method for detecting the lead position of a leaded component will be described below with reference to the drawings.

By setting a threshold value for the object imaged by the image pickup means (CCD camera) and binarizing it, the object is HIGH and the part other than the object is LOW.
And cut it. The part that changes from HIGH to LOW or from LOW to HIGH becomes the boundary between the object and the other part. Here, HIG of HIGH and LOW
The portion H is used as a boundary point as a point representing the boundary.

FIG. 5 is a diagram for explaining a conventional pattern recognition method. In the binary image of FIG. 5A, a HIGH point that changes from HIGH to LOW for the first time is detected as a boundary start point 19 by scanning in a predetermined direction from a point (nozzle center 9) inside the object. To do. A HIGH point that becomes a boundary in a predetermined direction (clockwise) is searched from the boundary start point 19 and the boundary start point 19 is again searched.
Find the sequence of boundary points until returning to.

From the boundary point sequence thus obtained, clockwise is "+180 degree corner 20" and "+90 degree corner 22", and counterclockwise is "-180 degree corner 21" and "-90 degree corner 23". By setting
Each corner can be detected as in (b). And, among these corners, +180 degree corner 2
Since 0 is the tip position of the lead, if the center point of this +180 degree corner 20 is detected, that point is the lead position 24.
Therefore, all lead positions 24 of the object can be detected.

[0006]

However, in the case of such a conventional pattern recognition method, the captured image must be binarized. When the boundary is traced, the boundary must be scanned in a fixed direction from the starting point 19 of the boundary, and finally all the boundary points of the object must be passed through to return to the starting point 19 of the first boundary point. In other words, it is a pattern recognition method under the condition that the outline of the object needs to be drawn with one stroke. In addition, as shown in FIG. 2, for an image in which the lead 7 portion is bright and the body 8 is dark depending on how to illuminate the object and the image capturing method, the lead 7 is HIGH and the body 8 and others are binarized. Is LOW
Therefore, there is a problem in that it is difficult to apply the conventional technique because the object cannot be represented by one boundary point sequence. Therefore, in order to solve this problem, the present invention aims to provide a pattern recognition method that can handle both a binary image and a grayscale image and that can reliably recognize even when the brightness is not uniform. is there.

[0007]

In order to achieve the above-mentioned object, the pattern recognition method of the present invention is a first method for detecting the inclination of an object with respect to an image obtained by imaging the object by the imaging means. A step, a second step of setting an area to be processed from the inclination of the object obtained in the first step, and a third step of projecting in a predetermined direction in the area to be processed, A fourth step of first-order differentiating the projection data obtained in the third step, a fifth step of determining the size of the object, and a first-order obtained in the fourth step. A sixth step of detecting a point at which the sign of the differential value changes to zero in a predetermined direction and becomes zero.

[0008]

According to the above method, there are the following two actions.

(1) It can handle binary images and grayscale (multivalued) images. (2) It is possible even if the object does not have uniform brightness.

[0010]

DESCRIPTION OF THE PREFERRED EMBODIMENTS A pattern recognition method according to an embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a flow chart of a pattern recognition method according to an embodiment of the present invention, and FIG. 2 is a diagram for explaining a method of detecting a lead position on one side of an object.

An inclination detection window 10 is set for a grayscale image (FIG. 2A) picked up by an image pickup means (CCD camera) from a given object size and nozzle center position, and this window is set. The inside is scanned from the direction of the arrow A, and the edge point 11 obtained first is detected for each line. After that, the edge point sequence is approximated by a straight line to obtain the inclination straight line 12 (step 1).

Next, the read detection window 13 is set according to the inclination of the inclination straight line 12 and the area is determined (step 2). This window is set so as to be parallel to the inclination straight line 12 and to include the lead, as shown in FIG.
It looks like (b). Then, by projecting the luminance value in the lead detection window 13 in a direction perpendicular to the inclination of the inclination straight line 12, data as shown in FIG. 2C can be obtained (step 3).

Furthermore, when the primary differentiation is applied to the data obtained by the projection (step 3) (step 4),
It becomes like FIG.2 (d).

Then, the center position of the lead is detected using the obtained data of the first differential 4. This detection method will be described below.

First, the primary differential value is scanned from the direction of arrow A, and the determination point A at which the differential value becomes larger than the positive threshold value 14 is obtained.
16 and a decision point B17 in which the differential value is smaller than the negative threshold value 15 are obtained. Then, the closest decision point A16
Then, a pair of determination points B17 is obtained, and all pairs whose width is smaller than the read width are extracted (step 5).

Finally, a coordinate (zero cross point 18) which becomes zero and changes from positive to negative between the determination point A16 and the determination point B17 is obtained. This zero-cross point indicates the center position of the lead (step 6).

By the above, the position detection of the lead facing in one direction is completed. For lead detection on the other side, refer to FIG.
As described above, the tilt detection window 10 may be set on each side to detect the lead position. Alternatively, as shown in FIG. 4, the lead position may be detected by setting the lead detection window 13 on each side from the obtained inclination straight line 12 and the size of the object.

[0019]

As described above, according to the pattern recognition method of the present invention, the following two effects can be obtained.

(1) It can handle binary images and grayscale (multivalued) images. (2) It is possible even if the object does not have uniform brightness.

[Brief description of drawings]

FIG. 1 is a flowchart showing an outline of a pattern recognition method according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating a method of detecting a lead position on one side of an object.

FIG. 3 is a diagram for explaining a method of detecting edges other than those detected in FIG.

FIG. 4 is a diagram for explaining a side detection method other than that detected in FIG.

FIG. 5 is a diagram for explaining a conventional pattern recognition method.

[Explanation of symbols]

 10 Tilt detection window 11 Edge point 12 Tilt line 13 Lead detection window 18 Zero cross point

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Takashi Shimizu 1006 Kadoma, Kadoma City, Osaka Prefecture Matsushita Electric Industrial Co., Ltd.

Claims (1)

[Claims] 1. A first step of detecting an inclination of an object with respect to an image obtained by imaging the object by an imaging means,
A second step of setting an area to be processed from the inclination of the object obtained in the first step, a third step of projecting in a predetermined direction in the area to be processed, and the third step. Of the first derivative of the projection data obtained in the step, a fifth step of determining the size of the object, and a fourth derivative of the first derivative obtained in the fourth step. A sixth step of detecting a point at which the code changes to zero in a predetermined direction and becomes zero.