Image analysis of Bi-discrete radial Cosine transform

Cosine transform seldom applies to pattern and object recognition due to its poor capabilities in describing image and calculating efficiency. In order to resolve these problems, a Bi-discrete radial cosine transform is proposed. The new cosine transform has two particular properties. First it takes Mukundan's square-to-circular transformation to project the square to the circle grids. Second different from classical radial orthogonal plane, a new discrete orthogonal polar plane is constructed by making the discrete cosine function orthogonal in radial direction to the discrete cosine function orthogonal in circumferential direction. Experimental results indicate that the proposed Cosine transform has good performance in image reconstruction and computing efficiency.