Asymmetric Stereoscopic Video Super-Resolution and Depth Estimation

Abstract: Recreation of full-resolution stereoscopic video from a asymmetric stereoscopic video is a challenging assignment. The existing techniques accept that the depth information is available, which forces an additional challenge in data acquisition. In this paper, we propose a novel plan that is fit for acquiring super-resolution and depth estimation at the same time from an asymmetric stereoscopic video. The proposed plot models the video super-resolution and stereo matching with a unified energy function. At that point, we apply alternating optimization technique to minimize this energy function, which can be implemented with a two-step algorithm. In the initial step, initial depth map is calculated using region based co-operative optimization technique to minimize this energy function, which can be implemented with two step function. In the second step, the super-resolution issue under the direction of the depth information is resolved. It is powerful on the grounds that each step provides extra benefit over the past step. These two steps are iteratively updated until stable depth and super-resolution results are acquired. Series of experiments are conducted on open stereoscopic video arrangements to assess the performance of the proposed technique. Based on comparison of objective indices and subjective visual results, it confirms that proposed scheme can achieve satisfactory super-resolution results and high-quality depth map simultaneously. Specifically, the subjective assessment experiment on 3D screen shows that this technique outperforms others and accomplishes the best visual sharpness.

Keywords: Asymmetric Stereoscopic Video, Super Resolution, Depth Estimation, Stereo Matching.