Edge-preserving directional regularization technique for disparity estimation of steroscopic images

Recently, due to the advances in visual technology, three dimensional(3D) i maging systems are becoming popular. One way of stimulating 3D perception i s to use stereo pairs, a pair of images of the same scene acquired from dif ferent perspectives. Since there is an inherent redundancy between the imag es of a stereo pair; data compression algorithms should be employed to effi ciently represent stereo pairs. Disparity estimation of stereoscopic image is an important stage for the stereo image compression. We propose a new di sparity estimation method using a directional regularization technique to p reserve edges well for stereoscopic image coding This method smooths dispar ity vectors in smooth regions and preserves edges ill object boundaries wit hout over-smoothing problem. The experimental results show that the propose d method achieves close matches between a left image and a right image as w ell as the improved coding efficiency. We use DPCM technique followed by th e Huffman entropy, coding for disparity map coding. In addition, the propos ed disparity estimation method can be applied to the high-quality intermedi ate view reconstruction.