A new block-matching algorithm based on subspace and partial distance search techniques in the wavelet domain

In this paper, a new fast block-matching algorithm for video coding is presented. In the algorithm, the partial distance search (PDS) technique is performed in the wavelet domain to eliminate the undesired blocks. Since the energy of blocks is concentrated in few coefficients in the wavelet domain, by starting the PDS at these coefficients, the algorithm can significantly reduce computational time for block matching. The PDS can be performed over the subspace in the wavelet domain to further lower the computational time at the expense of possible slight degradation in performance. In addition, the simple Haar wavelet can be used to remove the need for multiplication for blocking matching. Simulation results show that our algorithm is very effective for video coding where real-time processing is required.