CAJ | 학술논문

文章结合结构修复算法及纹理修复算法各自的优点，并考虑显著结构对图像修复的巨大影响，提出结合显著结构重构与纹理合成的图像修复算法。算法先利用形态学算子剥离待修复图像中细小结构与大块区域；然后利用快速结构修复算法对图像进行处理；再利用插值对待修复图像进行显著结构重构；最后利用基于改进优先级的加权匹配图像修复算法进行后续修复。实验结果表明，对既有显著结构同时又包含丰富纹理的待修复图像，与传统算法相比，本文的算法不但有更好的修复效果，而且耗时更少。
문장결합결구수복산법급문리수복산법각자적우점，병고필현저결구대도상수복적거대영향，제출결합현저결구중구여문리합성적도상수복산법。산법선이용형태학산자박리대수복도상중세소결구여대괴구역；연후이용쾌속결구수복산법대도상진행처리；재이용삽치대대수복도상진행현저결구중구；최후이용기우개진우선급적가권필배도상수복산법진행후속수복。실험결과표명，대기유현저결구동시우포함봉부문리적대수복도상，여전통산법상비，본문적산법불단유경호적수복효과，이차모시경소。
This paper presents an image restoration algorithm by integrating structure and texture features with reconstruction of significant structure of images. Firstly, the algorithm splits the fine structures and large areas in the damaged image using morphological operators. Secondly, the rapid structural restoration algorithm is used for image inpainting. Thirdly, significant structures in the image are reconstructed by interpolation. Finally, the large damaged areas are restored by the weighted matching image restoration algorithm based on the modified priority. For the damaged image with significant structure and texture, the experimental results show the better restoration result and less time consuming of the proposed algorithm than that of the traditional algorithms.