Image analysis by discrete radial Tchebichef moments

The feature extracted by the rotation invar -iants of radial Tchebichef moments could make the image analysis more effectively. However, the classical method adopted integral point sampling which has defects of too many sampling points in the centre of unit circle and insufficient samplings on the edge of the circle. It reduces the efficiency of feature extraction. In order to resolve this problem, we adopt Mukundan´s square-to-circular transformation to project square image to circle grids. By making the discrete radial Tchebichef polynomials orthogonal to discrete Fourier in circumferential direction, a kind of new discrete radial Tchebichef Fourier moments can be constructed. The experimental results show that the suggested method is better than Mukundan´s method as the size of image is large.