为了提高图像边缘结构信息的修复效果，本文优化的Criminisi算法优先权，侧重于提高数据项的鲁棒性，并在全局搜索寻找最佳匹配模块进行修复。且在修复结束后，对修复区域进行FDIM算法的无缝拼接处理，使图像的整体和谐统一。实验证明，该修复方法能够很好地处理边缘结构细节信息，并降低了错误匹配，满足人的视觉需求，具有相对较高的实用价值。

In order to improve the inpainting effect of marginal structure, the priority of optimized Criminisi algorithm focuses on improving the robustness of a data item in this paper, and then the optimized Criminisi algorithm searches for the best matching module in global image. After these, this paper uses FDIM which is a kind of seamless splicing algorithms in repair region, which makes a harmony of the whole image. Lots of experiments prove that this method can deal with the details of marginal structure effectively and reduce the inaccurate matching as well as meets the demand of human visual, so this method has relatively high practical value.