关键词: 数字图像修复, 像素相似, 域相似, 优先级

Abstract: Image inpainting based on partial differential equation and image completion based on texture synthesis is a very important method in digital image inpainting, in the present, these methods perform well, but they consume much time. This paper presents a new image inpainting algorithm based on the similarity of neighborhood. Firstly, the paper calculates the priority of every boundary point and makes the descending priority order as an inpainting order, the algorithm takes the similarity of pixel’s neighborhood to measure the degree of the similarity of pixel, it fully considers the influence of known information of neighborhood for pixel, then, calculates the average of the known pixels whose meet the requirements of similarity as the values to be filled point. Experimental results show that this algorithm can inpaint image better, and under the same scope of to be filled and effect of inpainting conditions, this algorithm has higher inpainting efficiency.

Key words: digital image inpainting, similar pixel, similar neighborhood, priority