Intermediate view synthesis from stereoscopic videoconference images

A procedure is described for stereoscopic videoconference system with viewpoint adaptation. The core to such a system is to synthesize the intermediate views from stereoscopic videoconference images with rather large baseline. The foreground object is first segmented by using intensity and disparity information. For this purpose, the region growing technique is used. The reliability of disparity estimation is then measured with a criterion based on uniqueness and smoothness constrains. In occluded areas and image points with unreliable disparity assignments, region-based interpolation strategy is applied to compensate the disparity values. Finally, an object-based interpolation algorithm is developed to synthesize arbitrary intermediate views. Experimental results with natural stereoscopic image pairs show that the proposed method can obtain the intermediate views with high quality.