光照不均匀图像的一种环带最优阈值分割方法

在水下探测、管道探测、工业内窥镜等视频监控系统中，南于主动补光，存在视频图像中央区域偏亮、四周偏暗的问题，无法直接进行图像分割。文章基于局部阈值化方法的思想，结合上述监控系统视频图像的特点，提出了环带划分的最优阈值分割方法。该方法的关键是根据图像亮区重心确定环带中心和根据目标尺寸确定环带宽度，进而把图像划分为若干子图像，然后对子图像运用局部阂值化方法寻找最佳阈值，实现图像分割。文中对水下鱼饵图像和字符探测图像分别运用全局阈值化方法、局部平均值阈值化方法及本文阈值化方法进行了仿真试验，实验表明本文方法对具有上述特点的图像的分割效果优于其他方法。

An Annulus Optimal Thresholding Method of Image in Nonuniform Illumination Conditions

In the video supervisory systems, such as underwater detecting, pipeline detecting or industrial endoscope, we can＇t directly use the classical thresholding method to segment the video images which are in nonuniform illumination conditions. According to the characteristics of the images, this article proposes an annulus dividing thresholding method based on the local thresholding thought. This method firstly divides an image into some sub- images, and then uses each local threshold value to segment the corresponding sub-image. The core of the method is to find the center of the bright image area and to set the size of the annulus. We do some simulation tests by the global thresholding method, local thresholding method and annulus dividing thresholding method. The experiment result shows that our method is better than others.