基于改进OTSU算法的红外图像分割

The Infrared Image Segmentation of Tank Based on Improved OTSU Method

红外图像分割是红外目标检测与识别领域研究中的一项重要内容。针对经典OTSU阈值分割算法无法适应目标模糊、对比度低的红外图像的缺陷,提出了一种改进的OTSU红外图像分割算法。该算法首先对红外图像进行预处理操作,再利用非线性平均灰度值的权重计算方法,克服了因目标靶面面积过小造成的阈值偏移。实验表明该改进方法较传统OTSU算法有更好的分割效果。

The infrared image segmentation belongs to the research field of infrared target detection and recognition. As the traditional OTSU threshold segmentation algorithm can’t work well with the infrared image which has low contrast and fuzzy target,one kind of improved OTSU algorithm was proposed. Firstly, the improved OTSU algorithm make a preprocessing of the infrared image,then using nonlinear average grey value of the weight calculation method to calculate weight. In this way,threshold error induced by small target area can be solved. The experiment results showed that the improved OTSU algorithm has a better performance than tradition OTSU threshold segmentation algorithm in the infrared image segmentation.