无形变视差调节快速算法

An undeformed fast algorithm for disparity adjustment

针对目前的3D播放器使用简便快捷的拉伸和平移算法实现深度感的调节产生明显的图像形变问题,本文提出了一种精确的视差调节模型,但模型调节视差需要获取原始深度值,计算复杂度高,导致难以流畅播放高分辨率3D视频图像。为此,进一步引入最小形变可察觉范围,通过假设原始深度的统计模型,把形变大小控制在最小形变可察觉范围内,在复杂度与形变程度上作出适当的平衡。VLC播放器实验表明,相比传统方法,本文算法形变程度降低超过50%,立体视觉主观效果更好;对于高清视频,算法效率提高48%。

In order to maintain and improve the visual experience and comfort of stereoscopic video, dis- parity adjustment becomes one of the key technologies of 3D applications. To solve the deformation prob- lems of the traditional algorithm, this paper proposes a precise parallax regulation model. Through affine transformation method,the model solves the problem of parallax adjustment after the object deforma- tion. But this model needs to get the original depth value to adjust the parallax, and the high computa- tional complexity leads to hardly playing high resolution 3D video images fluently. Therefore,we further introduce the minimum deformation range to be detected. By assuming the original depth of the statistical model,the deformation size can be controlled below the minimum deformation, which makes a proper balance between complexity and deformation degree. Finally, the modeling analysis and VLC player in- tegrating experiments show that the proposed algorithm reduces the degree of deformation by more than 50 % compared with traditional methods, and has better subjective stereoscopic effects. For high-defini- tion videos,the efficiency of the algorithm is improved by 48%.