计算机工程与应用 ›› 2014, Vol. 50 ›› Issue (18): 167-170.

• 图形图像处理 • 上一篇    下一篇

基于改进Criminisi算法的图像修复

李爱菊,钮文良   

  1. 北京联合大学,北京 102200
  • 出版日期:2014-09-15 发布日期:2014-09-12

Image inpainting based on improved Criminisi algorithm

LI Aiju, NIU Wenliang   

  1. Beijing Union University, Beijing 102200, China
  • Online:2014-09-15 Published:2014-09-12

摘要: 针对Criminisi算法难以获得理想的修复效果,且存在修复时间过长等缺陷,提出一种改进Criminisi算法的图像修复算法。改进优先权计算方式找到最优待修复块,完善最优匹配块搜索策略,找到最优匹配块,采用新的置信值更新方式以获得更为理想修复效果,通过仿真实验测试算法性能,结果表明,相较于Criminisi算法,改进Criminisi算法不仅获得了较理想的图像修复效果,而且大幅度减少了修复时间,提高了图像修复的效果。

关键词: 图像修复, Criminisi算法, 优先级, 图像纹理

Abstract: A novel image inpainting method based on improved Criminisi algorithm to solve the defects of Criminisi algorithm which hardly yields unfavorable results and costs much inpainting time is proposed. The priority levels of the blocks are obtained and the optimal inpainted block is selected, and then the optimum matching block in the image is found by improving search method, and the confidence value is updated to obtain good inpainting results. The simulation experiments are carried out to test the performance. The experimental results show that compared with Criminisi algorithm, the proposed algorithm is effective in both guaranteeing the quality of repaired images and saving much more repairing time.

Key words: image inpainting, Criminisi algorithm, priority levels, image texture