Feature-Points Based Shape Matching

In this paper, we propose a shape contexts based method which incorporates appearance similarity term into the correspondence estimation to improve the performance of shape matching. The optimal correspondence result then can be acquired by balancing the cost of matching and appearance similarity. On the other hand, a feature-points based matching algorithm is also presented to reduce the search space and improve the efficiency on shape matching. The algorithm is tested on two famous databases, namely, the database of MPEG-7 CE-Shape-1 part B and the coil-100 database. We also evaluate this algorithm on some images with noisy background. The experimental results show that our approach is robust and efficient in images with noise.