一种胶片档案斑块破损掩膜提取算法

A Film Archive Plaque Breakage Mask Extraction Algorithm

针对胶片档案影像在尺寸大且斑块破损小又多的情况下,需要遍历整张图片才能提取出斑块破损掩膜的问题,提出了帧差法和视觉显著图法相结合的胶片档案斑块破损提取方法。先对输入的胶片档案影像利用帧差法提取出可疑区域,接下来只需对可疑区域像素点,运用视觉显著图方法再进行二重阈值判定,最后将所有像素点判定结果汇总便得到一张二值化的斑块破损掩膜图。实验结果表明:此算法无须计算整帧图像的视觉显著图,就能将破损斑块掩膜较为完整地提取出来,并能兼顾算法效率和误检率。

Aiming at the problem that the film archive image needs to traverse the entire picture to extract the plaque damage mask when the size is large and the plaque damage is small and many,a film archive plaque damage extraction method combining the frame difference method and the visual salient map method is proposed.First,the input film file image is extracted by the frame difference method,and then only the pixels in the suspicious area need to be judged by the visual salient map method,and finally the results of all pixel judgments are summarized to obtain a binary plaque damage mask map.The experimental results show that the algorithm can extract the broken plaque mask relatively completely without calculating the visual salient image of the whole frame image,and can take into account the algorithm efficiency and false detection rate.