JP3170912B2 - Image contour tracking method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image contour tracking method for extracting a characteristic point of an image by eliminating an adverse effect of a noise region existing at a periphery of a camera field of view.

[0002]

2. Description of the Related Art A contour tracing method of an image for obtaining a position, a shape, and the like of an inspection object is performed by observing the inspection object with a camera, capturing the image, and tracking a boundary between the image and the background in a visual field. Are extracted, and the shape, position, center of gravity, etc. of the inspection object are determined from the positions of the characteristic points. Next, a conventional contour tracking method will be described.

FIG. 6 shows an image 2a of the inspection object 2 captured inside the field of view 1 of the camera. The inspection object 2 is larger than the visual field 1, only a part of the inspection object 2 enters the visual field 1, and most of the inspection object 2 protrudes from the visual field 1. Also field of view 1
There is a noise region 3 on the periphery of. A start point 4 is set at an appropriate position in the visual field 1 (for example, at the center of the visual field 1). Then, as shown by an arrow in the figure, scanning is performed from this start point 4 in a predetermined direction (upward in this embodiment) to switch the luminance switching point 5, that is, the boundary between the image 2a of the inspection object 2 and the background. By detecting and tracing the contour while meandering in a predetermined direction (counterclockwise in this embodiment) from the switching point 5, the image 2a of the inspection object 2 is obtained.
The feature point (inflection point) K is extracted.

[0004]

In the example shown in FIG.
The feature point K of the image 2a was able to be extracted. However, there are cases where the feature points cannot be extracted by the conventional method. FIG. 7 shows such a case. As shown, the image 2a captured by the camera is considerably inclined. In this case, if the contour tracking is performed in a predetermined direction from the start point 4, the user gets lost in the noise area 3 at the point 6, and the tracking of the contour becomes impossible as shown by the arrow in the figure. However, there is a problem that the feature point K cannot be extracted forever. Also, in the case shown in FIG. 6, if the tracking direction is set so as to trace the contour in the clockwise direction from the switching point 5 as shown by the dashed arrow, the point 7
In this case, the user gets lost in the noise region 3 and the feature point K cannot be extracted forever. Such a problem is likely to occur when the setting of the position of the start point, the tracking direction (clockwise or counterclockwise), and the setting of the tracking distance are inappropriate. Therefore, in the conventional method, after predicting the orientation of the image of the inspection object captured in the field of view of the camera,
You have to set the starting point, tracking direction and tracking distance,
There was a problem that software design was troublesome.

SUMMARY OF THE INVENTION It is an object of the present invention to provide an image contour tracing method capable of reliably extracting a feature point of an image of a large object to be inspected that protrudes from the field of view of a camera.

[0006]

According to the present invention, a pseudo contour is set inside a noise region existing at the periphery of a visual field, and the boundary between the image of the inspection object and the background is traced within the pseudo contour. and per to extract the feature points, previously set
Scan in the specified direction from the specified start point and
Once the boundary of the landscape is detected,
When the pseudo contour is reached, the boundary between the pseudo contour and the image
And once again reaches the boundary between the image and the background,
And the border between the image and the background . In addition, the pseudo outline is made the same color or the same luminance as the background of the image so that the pseudo outline is not visible in the background.

[0007]

According to the above configuration, the contour is traced within the pseudo contour, and the noise does not get lost, so that the characteristic points of the image of the inspection object can be reliably extracted.

[0008]

Next, an embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a perspective view of an electronic component mounting apparatus.
In the present embodiment, a description will be given of an example of an outline tracking dimension of an image of an electronic component, which is used for checking the type of the electronic component and detecting a position shift when the electronic component is mounted on a printed board by an electronic component mounting apparatus. The rotary head 11 is provided with a plurality of heads 13 having nozzles 12 for vacuum-sucking the electronic components 2 along a circumferential direction thereof. The rotary head 11 is driven by the drive unit 14 and rotates in the direction of the arrow. A table 16 on which a parts feeder 15 is mounted is provided behind the rotary head 11, and an XY table 18 on which a printed circuit board 17 is mounted is provided in front of the table 16. A camera 19 for observing the electronic component 2 vacuum-adsorbed to the nozzle 12 is provided below the moving path of the head 13.

When the rotary head 11 rotates in the direction of the arrow, the head 13 moves. However, when the nozzle 12 moves up and down just above the parts feeder 15, the electronic components 2 of the parts feeder 15 are sucked by vacuum. Pick up. Next, when the head 13 moves above the camera 19, the electronic component 2 is observed by the camera 19, and a characteristic point of the electronic component 2 is detected by contour tracing described later. The type confirmation and the calculation of the displacement are performed by a computer (not shown). Next, when the head 13 moves above the printed board 17, the nozzle 12 moves up and down again, and the electronic component 2 is mounted on the printed board 17.

FIG. 2 is a block diagram of the electric circuit. The binary light / dark image data captured by the camera 19 is input to the frame memory 8. For example, the image is bright, its value is “1”, the background is dark, and its value is “0”. The frame memory 8 has a pseudo contour writing unit 9 and an image recognition unit 10.
Is connected. Next, the contour tracking method will be described.

FIG. 3 shows an image 2 a of the electronic component 2 captured in the field of view 1 of the camera 19. As described above, the noise region 3 exists at the periphery of the visual field 1. Therefore, a pseudo contour 23 similar to the visual field 1 is set inside the noise area 3. The pseudo outline 23 is written with the same color or the same luminance as the background ("black" in the present embodiment). Therefore, the pseudo outline 23 is visible in the image 2a, but is not visible in the background. Although the electronic component 2 is larger than the visual field 1 and protrudes outside the visual field 1, it is taken into the visual field 1 in a correct posture without being inclined.

Now, scanning is performed in a predetermined direction (upward in the present embodiment) from a preset start point 4. When a boundary between the image 2a and the background is detected at a point 24, the scanning is performed in a predetermined direction (in the present embodiment). Image 2a and background in the counterclockwise direction)
Keep track of the boundaries. And after passing the feature point K,
When the pseudo contour 23 is reached at the point 25, the boundary between the pseudo contour 23 and the image 2a is tracked. After turning at the point 26, when the boundary between the image 2a and the background (point 27) is reached again , It changed the direction because in the background false contour 23 does not appear in the counterclockwise direction as, arrow
Follow the border between the image 2a and the background again as indicated by the mark.
And return to the original position (point 24). By tracking the boundary including the feature point K in this way, the feature point K is extracted.

By setting the pseudo contour 23 inside the noise region 3 in this manner, the feature point K can be reliably extracted without getting lost in the noise region 3. FIG. 4 is a flowchart of such contour tracking. The turning points 25, 26, and 27 are also feature points, but feature points extracted on the pseudo contour 23 are unnecessary and are ignored.
Such an ignoring process can be easily performed by software.

FIG. 5 shows a case where the image 2a is inclined. Also in this case, points 28 to 23 and image 2a
After turning at points 29 and 30 and turning at the point 31 toward the original position and tracking including the feature point K, the feature point K can be reliably extracted.

[0016]

As described above, according to the present invention, a pseudo contour is set inside a noise region existing around the visual field, and the boundary between the image and the background is tracked in the pseudo contour to determine the characteristic points. Since the extraction is performed, the adverse effect of the noise region existing at the periphery of the visual field is eliminated, and the feature points can be reliably extracted. In addition, the start point and the scanning direction can be freely set since the boundary does not protrude from the pseudo contour and does not get lost in the noise area.

[Brief description of the drawings]

FIG. 1 is a perspective view of an electronic component mounting apparatus according to an embodiment of the present invention.

FIG. 2 is a block diagram of an electric circuit according to one embodiment of the present invention.

FIG. 3 is an image diagram of one embodiment of the present invention.

FIG. 4 is a flowchart of an embodiment of the present invention.

FIG. 5 is an image diagram of one embodiment of the present invention.

FIG. 6 is a conventional image diagram.

FIG. 7 is a conventional image diagram.

[Explanation of symbols]

 1 field of view 2 electronic components 2a image 3 noise area 19 camera 23 pseudo contour

Claims (2)

(57) [Claims] An image contour tracking method for capturing an image of an inspection object larger than a field of view of a camera into a camera, tracking a boundary between the image and a background within the field of view, and extracting a feature point of the image. Te, set the false contour inwardly of the noise region present in the periphery of the field of view, per to extract the feature point tracking boundary of the image and the background within the pseudo contour, preset
Image and background by scanning in a predetermined direction from the specified start point
Once the boundary of is detected, it will follow this boundary,
When the pseudo contour is reached, the boundary between the pseudo contour and the image is added.
Traces and reaches the boundary between the image and the background again.
An image contour tracking method, wherein a boundary between an image and a background is tracked . 2. The method according to claim 1, wherein the pseudo contour has the same color or the same brightness as the background of the image.