X光安检图像低对比度区域目标的分割与增强算法

Segmentation and enhancement algorithm of targets in low contrast area of X-ray security image

为了增强显示安检图像中一些如植物风险源等因处于低对比度区域而难以识别的目标,提高安检人员的识别准确率,提出了粗分割细搜索的模糊增强算法。粗分割细搜索阶段,通过使用OTSU与结合了均值偏移的区域生长算法对低对比度区域目标进行提取。模糊增强阶段,根据感兴趣区域中像素隶属于目标区域的程度进行增强。实验表明,该算法能将低对比度区域目标提取出来并实现目标的增强,同时不改变图像的其他区域。

Some targets in the security inspection image,such as plant risk sources,are difficult to identify because they are in the low contrast area.In order to enhance their display images and improve the recognition accuracy of security inspectors,a fuzzy enhancement algorithm based on coarse segmentation and fine search is proposed.In the coarse segmentation and fine search stage,the low contrast area targets are extracted by using OTSU and region growth algorithm combined with mean shift.In the fuzzy enhancement stage,the pixels in the region of interest are enhanced according to their affiliation with the target region.Experiments show that the algorithm can extract the target in the low contrast area and enhance the target without changing other areas of the image.