基于改进样本块的数字图像修复算法研究

针对已有的基于样本块的纹理合成修复算法存在修复误差累积高的问题，重点研究了一种改进的基于样本块的数字图像修复算法。通过块匹配法与边缘驱动填充顺序的结合、全局搜索与局部搜索的结合，充分发挥修补过程中填充优先权的作用，有效地修补了图像受损区域的纹理和结构信息。仿真实验结果测试表明，与已有的传统算法相比，受损图像的修补更加完善，修复的误差累积得到了较好的改善。

Research on Exemplar-Based Digital Improving Image Inpainting Algorithms

Due to the high cumulative error of texture synthesis repair algorithm based on the sample block at present, fo- cus on an improved digital image inpainting algorithm. By the combination of block matching method and edge driving fill- ing order,global searching and local area searching,giving fully filling priority in the repair process, repairing the texture and structure of the damaged area of the image efficiently. The results of simulation tests show that, compared with the existing traditional algorithm, the inpainting of the damaged image, accumulation of the repairing errors has been good im- provement.