基于深度图象Warping的动态图形反走样算法

时间域上的动态图形反走样算法是提高序列画面图象播放质量的关键技术。由于该技术需在时间域上加密采样,因而其计算过程非常耗时。本文首先给出了一种新的深度图象ｗａｒｐ－ｉｎｇ算法,并用它来快速生成由于摄像机参数变化而产生的静态景物画面的细微变化,而运动景物画面则根据其深度值复合到相应时刻的静态景物画面中。利用这一技术,我们可快速地在相邻帧画面中间进行加密采样。而Ｇａｕｓｓ点优化采样策略则极大地提高了反走样算法的效率。理论和实验结果表明,本算法能高效地实现复杂场景图形的动态反走样,它不仅独立于场景复杂性,而且可根据画面质量要求来调整精度。

Anti Aliasing for Animation Sequences Based on the Warping Technique of Range Images

The temporal anti aliasing algorithm is the key to improve the quality of animation sequences. Since this technique needs super sampling over the time domain, it is very time consuming. This paper first presents a new warping algorithm, with which we can quickly generate the minor change of static images with the change of the camera's parameters. The moving objects are then composed to the corresponding images of static objects according to their depth information. Therefore, the sampling images between successive images can be quickly generated. Meanwhile, Gauss sampling technique greatly improves the efficiency of our anti-aliasing algorithm. Both theoretical and experimental results demonstrate that the algorithm can be implemented effectively.