Abstract: A mesh image distortion algorithm based on movement decomposition of feature point and scattered data interpolation was presented，to improve the performance of traditional 2-pass scan mesh distortion method，which has disadvantages on scan sequence and effect of distortion.Pixel was mapped from original image to target image，then use scattered data interpolation algorithm to the scattered point in order to resume the pixel information of target image.To improve the efficiency of grey mapping，we presented a triangle linear interpolation based on Delaunay triangulation to deal with large-scale scattered data.Finally an application instance was given to verify the desirable results of this method compared with the traditional one.

Key words: mesh distortion, image distortion, grey mapping, scattered data interpolation

摘要： 提出一种基于特征点运动分解和散乱灰度数据插值的网格图象变形算法，以改进传统的两步扫描网格变形法在扫描顺序和变形效果上的不足。将原始图象的象素坐标一次性映射至目标图象，再对映射后得到的散乱坐标点的灰度进行散乱数据插值以恢复目标图象的象素信息。为了提高灰度映射的效率，引入一种基于Delaunay三角剖分的三角线性插值的方法来处理大规模散乱数据的插值。最后通过实例证明该算法的变形效果较两步扫描网格变形法有显著提高。

关键词: 网格变形, 图象变形, 灰度映射, 散乱数据插值