红外技术, 2017, 39 (3): 221, 网络出版: 2017-04-10
基于蚁群算法的 Criminisi图像修复
The Criminisi Algorithm Based on Ant Colony Optimization for Image Inpainting
摘要
提出了一种基于蚁群算法的 Criminisi图像修复算法,将蚁群算法应用到 Criminisi图像修复算法的最佳匹配模板搜索中。首先计算待修复区域优先权;然后蚁群寻找搜索路径中留下的信息素,沿着信息素最多的路径寻找到最佳匹配模板;最后更新置信度,直到修复结束。实验结果表明,修复后的图像 PSNR较高不易陷入局部最优,能较快速地搜索到最佳匹配模板。
Abstract
An Criminisi image inpaiting algorithm based on ant colony algorithm is proposed, which applies Ant colony algorithm to search the best matching template in Criminisi image inpaiting algorithm. Firstly, the priority area to be inpainted is calculated. Then, ant colony finds the pheromoneleft in searching path, and follows the path with most pheromones to find the best matching template. Finally, the confidence is updated until the end of inpaiting. Experiments show that the PSNR value of the image after inpaiting is higher and local optimum is avoided. The method can more quickly search for the best matching template.
郑玉婷, 吴谨. 基于蚁群算法的 Criminisi图像修复[J]. 红外技术, 2017, 39(3): 221. ZHENG Yuting, WU Jin. The Criminisi Algorithm Based on Ant Colony Optimization for Image Inpainting[J]. Infrared Technology, 2017, 39(3): 221.