具有规则度约束的红外图像多层阈值分割方法

Infrared Image Segmentation Using Multi-threshold and Regulation

针对红外图像的噪声和模糊边缘给红外图像的分割和目标检测带来的困难,为了得到边缘连续的目标区域,提出了一种具有规则度约束的多层最佳阈值图像分割方法。在根据灰度对图像进行多阈值的初始分割的基础上,通过各个区域的规则度等参数对分割区域进行过滤处理,消除过分割区域,降低因为过度分割造成的目标识别困难。实验结果表明,该算法具有良好的效果和实用价值。不仅用于红外图像,也可以应用于自然光图像的分割。

In the infrared images, segmentation and objective detection are difficult because noises and blurry edges. For obtaining objective regions with continues boundary, image segmentation method based on multi-best-threshold and regulation is presented. At first, according to the multi-best-threshold to segment the image and obtain initial segmented regions, then it filters segmented regions by the regulation to eliminate over-segmented regions for avoiding the recognition difficulty. The experimental results show that the algorithm is effective and has practical value. Not only used in the infrared images, and also can be applied in the images of natural light.