This paper presents a new stereo matching algorithm for intermediate view reconstruction of stereoscopic television images. The proposed stereo matching algorithm reduces the computational burden of disparity estimation by assuming that the disparity is constant over blocks of NtimesN pixels. The disparity per block is hierarchically estimated. A cost function deduced from maximum a posteriori disparity estimation is taken as a block similarity measurement for matching. To minimize the cost, an algorithm using a dynamic programming technique is proposed. The optimization algorithm considers the costs of N+2 possible nearest neighboring candidate block pairs, which have a maximum disparity difference of N pixels. Experimental results obtained with test image pairs show that a block size of 4times4 pixels was found to be the best for the image spatial resolution tested in this paper. Given this block size of 4times4 pixels, the computational burden of the proposed algorithm can be reduced by as much as 89%, compared to a reference algorithm that computes the disparity per pixel, without sacrificing picture quality