关键词: 图像修复, 样本块修复, 算法改进, 眼肌图像

Abstract: Abstract: Because of moisture in pork, there would be reflective area in the image that took under illumination, which lead to the loss of image information．In order to improve the accuracy of related feature parameters in obtaining image，it is necessary to restore the reflective area. Image pre-processing and image inpainting technology was used to restore the image. According to the color feature of the reflection area, a series of pre-processing method was used to position and obtain the reflection area. Results of the restoration in image reflection area was compared in classical image inpainting method, inpainting algorithm based on exemplar patch was taken as the best inpainting method. Inpainting algorithm based on exemplar patch was improved, method of restoring the reflection area in sequence was used to reduce the calculate frequency of edge contour priority, in the same time, by using twice matching method，the amount of scanning data was reduced in the process of searching the best matched block，and the repair time was saved about 3/4 of original algorithm．The experimental results show that the improved inpainting method is able to restore the information from reflection area in image, and use less time compared with original algorithm.

Key words: image inpainting, Criminisi algorithm, algorithm improvement, rib-eye image