在经典 Criminisi 图像修复算法框架的基础上，针对优先权可靠性低和全局搜索最佳模板效率低、错误匹配率大的缺点进行改进。改进的算法为基于萤火虫算法(FA)的快速 Criminisi 图像修复算法。首先从数学的角度引入正规化函数至置信度，以此提升优先权计算的可靠性；然后引入 FA 到最佳模板的搜索与填充中，能够有效地将全局搜索与局部搜索有效地结合，鲁棒性较高，提高效率且错误匹配率低。实验结果表明：采用本文的改进算法能在保证修复质量的基础上降低时耗，提高效率。
On the basis of Criminisi algorithmic framework,a fast Criminisi image inpainting based on firefly algorithm(FA)is proposed for the deficiencies of the priority and the best match template. Firstly,normalization function is introduced to the confidence to improve the reliability of the priori-ty.Then,in order to improve and reduce error rate,FA which can combines global search and local search effectively is introduced to search the best template.Experimental results show that on the ba-sis of guarantee the quality of image inpaiting,the improved algorithm can reduce the cost and im-prove efficiency.
参考文献
| [1] | Bertalmio M;Sapiro G;Caselles V.Image inpainting[A].USA:John Seely Brown,2000:417-424. |
| [2] | 林云莉;赵俊红;朱学峰.基于图像分解的图像修复技术[J].计算机工程,201036(10):187-192. |
| [3] | 朱文浩;魏国刚.基于样本的纹理合成[J].中国图形图像学报,200813(11):2063-2069. |
| [4] | Criminisi A;Perez P;Toyama K.Region filling and object removal by exemplar based inpainting[J].IEEE Transactions on Image Processing,200413(09):1200-1212. |
| [5] | 郭勇;王梅.基于改进样本块的数字图像修复算法研究[J].软件导刊,201312(10):156-158. |
| [6] | 常晨;尹立新;方宝龙.一种改进的 Criminisi 图像修复算法[J].计算机应用与软件,201229(09):238-267. |
| [7] | 姚建亮;彭宏京.一种改进的基于样图的图像修复法[J].电子科技,201023(01):100-103. |
| [8] | 刘洋;王昊京;田小建.采用区域分割的变尺寸样本块高效图像修复[J].光学精密工程,201018(12):2657-2664. |
| [9] | 刘长平;叶春明.一种新颖的仿生群智能优化算法:萤火虫算法术[J].计算机应用研究,201128(09):3295-3297. |
- 下载量()
- 访问量()
- 您的评分：
-
10%
-
20%
-
30%
-
40%
-
50%