On Vision-Based Orientation Detection Analysis of Industrial Object with Circular-Shape Features

For the need of accuracy orientation location about work piece with circle features in vision-based automatic assemble industry, a kind of orientation detection method about work piece based on monocular vision was presented to locate the work piece and then adjust the orientation of assembly parts to guarantee the assembly precision. Ellipse parameters and center point of contours about holes was solved by extracting the center and edge about given the work piece with four holes in the experiment, orientation and location about surface of work piece can be derived. And then Automatic assemble was implemented by orientation of work piece feeding back to robot manipulator to adjust the orientation of assembling part. Experiment analysis has demonstrated that the uncertain of position is less than 8 um and the uncertain of orientation is less than 5 degree.