High Resolution Centroid Hirschman Descriptor For Moving Object Detection
Xi Peng, Victor DeBrunner · 2018 52nd Asilomar Conference on Signals, Systems, and Computers · 2018
In applications where one is attempting to detect an object in an image, the centroid Fourier descriptor (CFD) is often used because it is invariant to the starting point, scale invariant, and robust to noise. It is widely used in shape identification. However, the CFD is not of sufficient resolution to distinguish very similar shapes. In this paper, we introduce a high-resolution centroid shape descriptor based on the Hirschman Optimal Transform (HOT). Because of its relationship to the DFT, our newly proposed algorithm will inherit the invariances as well as the robust noise performance. However, when compared to the CFD, our centroid Hirschman descriptor (CHD) is superior both in its computational efficiency and its ability to recognize small changes in the object shape. We thus believe that the CHD is a better choice for detecting object movements than is the CFD.