In this paper, we propose an improved image in-painting method based on Fast Matching Method (FMM) algorithm. The traditional approach speeds less time but it cannot contribute an optimal edge result. To overcome this disadvantage and improve the edge effect. First we use gradient matrix to select less but more significant pixels to join into the gray value calculation. Secondly we use an edge prediction method to predict the edge in the in-painting region and reset the in-painting sequence. Furthermore, this procedure also had an advantage in in-painting the image which had a large destroyed region. Therefore, our improved method contributes an obvious edge for in-painting procedure than the traditional method.
展开▼
