拓扑梯度耦合FCMC的全自动图像修复优化算法

目的当前图像修复算法的损坏区域大都是依靠人工来确定,难以自动鉴定损坏区域,使其修复效率较低。此类算法通过利用像素缺失区域的间断边缘来完成填充,导致重构图像视觉间断,且都是依赖随机修复路径,增加了算法时耗。提出拓扑梯度最小重构路径耦合FCMC(Fuzzy C-mean Clustering)的全自动图像修复算法。方法基于图像损坏区域与完好区域之间的性质差异,引入模糊C均值(FCMC),通过损坏区域的聚类中心与各像素之间的距离来计算隶属度函数,设计基于FCMC的损坏区域自动鉴定算法,以自动识别待修复区域;再嵌入拓扑梯度,定义像素缺失区域的关键点选择规则,建立权重距离函数,得到像素缺失区域的连续轮廓,设计最低修复路径成本方案,完成图像重构;以PSNR(Peak Signal to Noise Ratio)为评估指标,构造图像修复反馈机制,优化修复图像。结果仿真结果显示:与当前图像修复算法相比,该算法可自动鉴定图像像素缺失区域,能够提取像素缺失区域的连续轮廓。同时,具有更好的修复视觉效果与更高的修复效率,重构图像不存在模糊与视觉不连通。结论提出的算法能够实现图像的全自动修复,可提高修复图像质量与效率。

The Automatic Image Inpainting Optimization Algorithm Based on Topological Gradient Coupled FCMC

Objective The damaged regions of current image inpainting algorithms are mainly determined manually, and cannot identify the damaged regions automatically, leading to low repairing efficiency. This type of algorithms repairs the image by extracting the non-continuous edge of the pixel missing regions, which often causes visual discontinuity of the reconstructed image, and the inpainting path is always randomly determined, which increases the time consumption of the algorithm. In this paper, an automatic image inpainting algorithm based on topological gradient minimum inpainting paths coupled fuzzy C-mean was proposed. Methods The goal of automatically identifying the damage area was achieved by introducing the fuzzy C-mean to compute the membership function of the distance between pixels and cluster center in the damaged regions. Then the continuous edge of the pixel missing region was obtained by embedding topological gradient, defining the key point selection law of the pixel missing region and establishing weight distance function; the reconstruction path with lowest inpainting path cost was designed and the image was reconstructed. Finally, the image inpainting feedback mechanism was constructed based on PSNR to optimize the inpainting image. Results The simulation results showed that comparing with other image inpainting algorithms, the proposed algorithm could automatically identify the pixel missing region and extract the continuous contour of the pixel missing region. Meanwhile, this inpainting algorithm had a better inpainting effect and higher efficiency to eliminate the blurring effects and visual discontinuity. Conclusion The proposed algorithm could realize automatic image inpainting, and could improve the quality and efficiency of the repaired images.