非均匀纹理图像大区域修复算法

为了改善实际非均匀纹理图像大区域修复效果,提出了一种改进的纹理合成算法.在传统图像修复算法优先权系数的基础上,增加了方向性优先权系数,为纹理合成时各点的传播方向和进度提供索引;同时,针对非均匀纹理图像渐近变化的特点,将以待修复块为中心的扇形区域作为最优匹配块的搜索区域,以减少误匹配.仿真实验结果表明,该方法能够有效克服传统纹理合成方法没有考虑方向性的缺点,对实际大区域非均匀纹理图像取得了较好的修复效果.

Non-Regular Texture Image Completion Algorithm in Large Region

In order to improve the completion effect of actual non-regular texture images in large region,an im-proved texture synthesis algorithm was proposed.Based on the traditional priority coefficient,the completion algo-rithm was improved by adding a directional priority coefficient to index the texture propagation direction and process of each point.Moreover,in order to reduce the mismatch,the patch-centered sector was taken as the searching scope for the best matching sample,taking into account the gradual changing characteristic of non-regular texture images.The results of the simulation experiment show that the algorithm can effectively overcome the directional problem,which was not covered by traditional texture synthesis methods,and get perfect completion effect on actual non-regular texture images in large region.