Abstract: To solve the problem of lost information of scanned core image, an inpainting algorithm based on total variation model was presented. Using the information of the reference pixels in the neighborhood area of the inpainted area, the information was diffused to the inpainted area. The neighborhood relativity coefficient was introduced weighing the influence on target pixel point by inpainting neighborhood boundary. The experimental results show that the new method improves the inpainting effect, and is effective for core image inpainting.

Key words: total variation, core image inpainting, neighborhood relativity coefficient

摘要： 针对岩心扫描图像信息缺失的修复问题，提出了基于整体变分模型的修复算法。利用图像待修复区域邻域的参考像素信息,从待修复区域边缘逐步向待修复区域内部扩散，同时采用了邻域相关系数来衡量待修复区域邻域边界对目标像素点的影响程度,对算法进行了改进。通过仿真实验表明，改进后的算法与原方法相比，修复效果得到了改善，可以有效完成对于岩心图像的修复。

关键词: 整体变分, 岩心图像修复, 邻域相关系数