In a robot assembly operation, the information of dimension such as the center position and the diameter is necessary. But it is impossible to recognize 3-D coordinate from single camera due to projection.
The purpose of this thesis is to present a method to obtain the necessary 3-D information of given object through 2-D image processing by using single camera.
The methods employed for recognition include a edge detection, the edge tracking with thinning on 2 and 1/2-D objects. For restoration 3-D real coordinate from 2-D image, at first, the height of object should be obtaind. And the height of object enables us to reduce the 3-D problem to 2-D problem on 2 and 1/2-D objects. Thus we seek all data for assembling task.
The proposed method is verified good for obtaining position data. The whole system is composed of vision system and robot system for peg-in-a-hole task.