Rigid and non-rigid object image matching using deformable object image discrimination

This paper proposes the image matching method that can match rigid object image and non-rigid object image by utilizing the same feature. To this end, first determines the matching of a rigid object image through geometric verification and then discriminate the non-rigid deformable image from the verified result by using supervised learning. Lastly, this paper proposes the method to match a non-rigid object image through clustering of feature matching-pairs in relation to the discriminated result. This paper confirmed that the proposed method had a lower time complexity and a higher matching success rate and accuracy than the conventional method.