A Shape Recognition Algorithm Robust to Occlusion: Analysis and Performance Comparison

This paper introduces a new approach for recognizing two-dimensional shapes called the SKS algorithm and compares it with three other state-of-art methods in detail. These include the Hu Moments, CSS matching and Shape context. The algorithm uses the philosophy of evidence accumulation similar to generalized Hough transform and is highly parallel in nature. The performance of each algorithm is evaluated under affine transforms translation, rotation in the plane, scale(zoom) and also under partial occlusion.