Optimal Frequency SSD Matching Using Cosine Kernel

Fast robust correlation is a recently published correlation which makes use of the cosine function. It works by expressing any robust matching criteria in terms of cosine series. In this paper we show the high relationship between the principle of the fast robust correlation approach and the SSD metric. We demonstrate that using Taylor Lagrange approximation of cosine function we can express the well known sum square difference in terms of one cross correlation operation using the same principle of the fast robust correlation. Speed is obtained from computing the correlation in the frequency domain.