一种新的基于破损区域分块划分的图像修复算法

针对破损区域较大、结构信息复杂的图像修复难题,在分析手工修复方法的基础上,提出一种基于破损区域分块划分的图像修复算法。根据边界线破损而断裂的图像,先估算破损区域边缘各断裂边界线的走向;再将同一条边界线的各断裂边界线进行平滑连接,可将破损区域划分成不同的块;然后分别计算各块的优先等级;并利用BSCB算法,按优先级对各块进行修复;最后,将算法与BSCB算法及近年来提出的图像修复算法进行比较。结果表明,所采用的图像修复算法修复结果图像结构信息更清晰、图像边界过渡更自然、更符合视觉感知;并具有更高的峰值信噪比。

A novel image inpainting algorithm based on partition block of damaged region

According to the problems existed in image inpainting with larger damaged region and more complex structure information, combined with the manual repair method. This paper presents image inpainting algorithm based on partition block of damaged region. Damaged Boundary line of image trend is estimated at first, and then the line which belongs to the same damaged boundary region is connected smoothly, the damaged region is divided into small partition blocks; the priority of those blocks is calculated respectively, BSCB algorithm is employed to inpainting the damaged blocks with the different priority; By comparison, BSCB algorithm and the latest introduced image inpainting algorithm are applied to image inpainting. The comparison results have shown that the proposed algorithm is effective and feasible in image inpainting of large damage region and more complex structure, and have better inpainting ability, obviously the inpainting results are more suitable to the visual perception, which also have high PSNR.