A Polygon Detection Algorithm for Robot Visual Servoing

The problem of detecting polygon structures in images arises in many applications, from building extraction to cryo-electron microscopy. In this paper, A polygon detection algorithm (PDA) based on line detection is proposed. First of all, we introduce a parallel thinning algorithm to eliminate useless feature points. Then the feature points are organized as disjoint feature contours. For each independent contour, we detect line segment candidates based on RANSAC algorithm. With these candidates, a polygon (triangle, rectangle or other polygons) is composed. Compared to the state-of-the-art shape detection algorithms, PDA is much faster and it can detect any polygons such as triangles, rectangles, etc. We introduce PDA to a 6DOFs robot visual servoing platform and the experimental results show that significant improvements in time efficiency and performance robustness of the proposed algorithm have been achieved.