一种交互式的岩心图像修补方法

An Interactive Core Image Inpainting Method

传统的岩心图像修补方法往往无法对不同宽度的待修补域达到理想效果,并且难以对待修补区域做出选择,这在实际应用中很难满足需要。针对这些问题,提出基于联合一阶与二阶全变差(Combined TV-TV2)的交互式岩心修补方法。该方法首先进行均值漂移对图像进行预分割,然后使用基于区域合并的交互式最大相似度算法确定待修补的区域,最后以联合TV-TV2修补模型为基础,应用分裂Bregman迭代进行求解。实验结果表明,待修补的区域能够被确定,而对于不同宽度的修补区域,该算法能够获得满意的修补效果。

Traditional core image inpainting methods often cannot achieve desired results on different widths of inpainting fields, and cannot determine the inpainting domain, which in practice is difficult to meet the needs. For these problems, propose an interactive core inpainting method based on combined TV-TV2 model. First,get a pre-segmentation result by mean-shift. Then use region merging based interactive maximum similarity algorithm to determine the area to be repaired. Finally, according to combined TV-TV2 model, apply split Bregman iteration to solve the problem. Experimental results show that the inpainting domain can be precisely determined,and for different widths of the inpaiting domain, the method can obtain satisfactory result.