Depth-Image Based View Synthesis for Three-dimensional Television

摘要

The trend in three-dimensional television (3DTV) applications demonstrates the interest in virtual view synthesis techniques. A depth-image based view synthesis method which utilizes a main color image with associated depth map and an assistant color image is presented in this paper. The depth-image based view synthesis method is performed through extracting essential information from an assistant color image to fill the exposed holes and wrongly mapped pixels in the image warped from a main color image on the basis of depth-image-based rendering (DIBR). During the synthesis, the main color image takes the leading position and a differential distance-and-depth weighted (DDW) algorithm is also proposed to improve the DIBR performance. Experiments results are presented to demonstrate the effectiveness of the proposed method through assessment of image quality and algorithm cost.