由于肉品中存在水分，在可见光下采集图像就会产生反光区域，进而导致肌肉信息部分丢失。为了提高获取肉品图像相关信息的准确性，需尽量还原反光区域。在基于样本块的图像修补算法基础上，结合肉品图像特点，提出改进算法：用Prewitt算子代替Robert算子,使信息缺损区域的线性结构得到更好的保持；采用二次匹配法减少搜索最佳匹配块过程中的扫描数据量，比原算法节约约2/3的修复时间；在修复过程中加入α剪裁均值滤波消除接缝效应，使得修复结果更好地满足视觉心理学要求。实验结果表明，改进的修补算法在效率和修复效果上优于原算法，不仅可用于肉品图像反光区域的修复，对其他图像的修复也有效。

Because of moisture in meat, there would be reflective area in the meat image under visible light, which lead to the loss of some muscle information. In order to improve the accuracy of obtaining meat image’s related information, it is necessary to restore the reflective area as much as possible. Based on the algorithm of image completion by exemplar path, with the feature of meat image, an improved algorithm is presented in this paper. First, replacing Robert operator with Prewitt, so that the LISREL of information lost area get better preserved. Then, by using twice matching method, the amount of scanning data is reduced in the process of searching the best matched block, and the repair time is saved about 2/3 of original algorithm. After that, to eliminate seam effects, adding α-cut Mean Filter in the process of inpainting, it makes the repair results meet the requirements of visual psychology better. The experimental results show that both the inpainting efficiency and effect of improved algorithm are better than the original, and the method can not only repair the reflective area of meat image, but also be useful for other images’ inpainting.