划分特征子区域的图像修复算法

发布时间：2018-07-27 16:33

【摘要】：为了解决含有丰富纹理信息和复杂结构信息的大破损区域中的缺失信息修复的问题,提出了一种划分特征子区域的图像修复算法。首先,根据图像中包含的不同特征,运用特征公式进行特征提取,再通过统计特征值划分特征子区域,提高了图像修复的速度;其次,在原Criminisi算法的基础上改进了优先级的计算,通过增大结构项的影响,避免结构断裂的产生;然后,通过目标块和其最佳邻域相似块共同约束样本块的选取,确定最佳样本块集;最后,利用权值分配法合成最佳样本块。实验结果表明,所提算法相比原Criminisi算法,其峰值信噪比(PSNR)提升了2~3 d B,相比基于稀疏表示的块优先权值计算的算法,其修复效率有明显的提高。所提算法不但适用于一般小尺度的破损图像的修复,而且对于含有丰富纹理信息和复杂结构信息的大破损图像的修复效果也更佳,并且修复后的图像更加符合人们视觉上的连通性。
[Abstract]:In order to solve the problem of repairing the missing information in large damaged regions with rich texture information and complex structure information, an image restoration algorithm is proposed to divide the feature sub-regions. First of all, according to the different features contained in the image, the feature formula is used to extract the feature, and then the statistical eigenvalue is used to divide the feature sub-region to improve the speed of image restoration. On the basis of the original Criminisi algorithm, the priority calculation is improved, and the structure fracture is avoided by increasing the influence of the structure item, and then the optimal sample block set is determined by the selection of the sample block constrained by the target block and its optimal neighborhood similar block. Finally, the weight distribution method is used to synthesize the best sample block. The experimental results show that compared with the original Criminisi algorithm, the proposed algorithm improves the PSNR (PSNR) by 2dB, and improves the repair efficiency of the proposed algorithm compared with the algorithm based on sparse representation of block priority value. The proposed algorithm is not only suitable for repairing small scale damaged images, but also better for large damaged images with rich texture information and complex structure information. And the restored images are more in line with the visual connectivity of people.
【作者单位】：西南交通大学信息科学与技术学院;西藏大学工学院;
【基金】：国家自然科学基金资助项目(61461048)~~
【分类号】：TP391.41

【相似文献】