Segmentation-based adaptive vergence control for parallel multiview stereoscopic images

The vergence control is important for multiview stereoscopic images captured by a camera array to acquire natural and clear stereoscopic effects. In this paper, we derive the basic characteristics of vergence control for parallel multiview stereoscopic images. Then we segment the stereoscopic images based on Mean Shift, and shift the images by the disparity of the object in the central region to achieve the adaptive vergence control. We have evaluated our method on Middlebury data sets and artificially synthesized images. Experimental results show that the proposed method is effective, and the generated stereoscopic images could reproduce the 3D scenes vividly in a multiview autostereoscopic display.