基于像素权值的高效小波图像修复算法  

Efficient Wavelet Image Inpainting Algorithm Based on Pixel Weight Values

作　　者：吕永利[1] 姜斌[1] 包建荣[1] 

机构地区：[1]杭州电子科技大学通信工程学院,浙江杭州310018

出　　处：《信息与控制》2015年第1期104-109,共6页Information and Control

基　　金：国家自然科学基金资助项目(61001133);浙江省自然科学基金重点资助项目(LZ14F010003);浙江省数据存储传输及应用技术研究重点实验室(杭州电子科技大学)资助项目

摘　　要：针对传统图像修复算法计算量大、修复耗时较长、复杂度高等缺点,提出了一种基于像素权值的高效小波图像修复算法.该算法先对受损图像进行小波分解,再快速定位待修复区域,后根据待修复区域及其邻域像素值计算相应像素权值,并用计算所得像素权值及邻域内已知像素值完成受损像素点的修复,最后由小波重构得到修复后图像.仿真表明:在类似修复图像视觉效果前提下,该算法执行速度比传统修复算法有较大提高,复杂度也有较大下降,适用于高效实时图像修复.Traditional image inpainting algorithms suffer from large computational costs, duration of run, and com- plexity. This paper proposes an efficient wavelet image inpainting algorithm based on pixel weight to overcome these challenges. First, the damaged image is decomposed by wavelet transformation and the regions to be in- painted are quickly located. Then, the pixel weights are calculated according to both the regions to be in- painted and the adjacent pixels around it. Subsequently, the calculated pixel weights and the known neigh- borhood pixels are used to finish the inpalnting of the damaged pixels. Finally, the inpainted image is ob- tained by wavelet reconstruction. Simulation results show that the speed of the proposed algorithm is remark- ably faster than that of the traditional inpainting algorithm, and the complexity also significantly decreases, under similar inpainted image visualization effect. Therefore, the proposed algorithm is very suitable for use in efficient real-time image inpainting scenarios.

关 键 词：图像修复 小波分解 像素权值 小波重构 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]