An Advanced Fourier Descriptor Based on Centroid Contour Distances

In order to distinguish two shapes with similar centroid contour distances. This paper proposes a novel descriptor-AFD-CCD (advanced Fourier descriptor based on centroid contour distances), which is based on FD-CCD ( Fourier descriptor based on centroid contour distance ) descriptor. FD-CCD only provides the distance information from the centroid point to the contour point, and the lack of direction information affects the accuracy of matching. The algorithm uses circularity and rectangularity to supplement discriminability, with low computational complexity. Tests were performed on three datasets. The experimental results show that AFD-CCD has further improved the matching accuracy while ensuring that the matching speed does not decrease.