摘要
This paper introduces an algorithm for 2D-to-3D video conversion, together with the corresponding real-time implementation on mobile devices. Saliency based on color histogram is used to build depth map in this algorithm. To eliminate the instability problem of generated saliency, a low-complexity and low-latency scene-detection method is adopted taking advantage of color histogram produced before. The experimental results show that the algorithm is able to create depth maps of good equality in various scenes in real-time and effectively alleviate the depth instability problem(1).
- 出版日期2015-11
- 单位清华大学