基于简化Mumford-Shah模型的水平集红外图像分割算法

由于红外图像大多具有目标模糊,对比度低的特点,传统的分割方法容易受到噪声和边界轮廓的影响而导致分割效果不佳,提出了一种基于简化Mumford-Shah模型的水平集红外图像分割算法.该算法能够通过将初始闭合曲线嵌入水平集函数,利用函数的求解从而达到图像分割的目的.仿真实验结果表明,该分割算法与初始轮廓线位置无关,受边界轮廓线和图像噪声的影响较小,具有较强的鲁棒性,在目标与背景灰度级差别较小的红外图像的分割中取得了较好的效果.

Level set infrared image segmentation based on simple Mumford-Shah model

The infrared image has the characteristic of objective fuzz. A level set PDE based on simplified Mumford-Shah model for image segmentation is proposed. It can overcome the deficiency of conventional methods for infrared image segmentation which is influenced by image noise and edge contour. Infrared image segmentation is accomplished by embedded initial curve in the level set function and obtained the solution of function. The simulation proves the method is a robust approach to infrared image which is independent of the initial curve and image noise. This method has a good segment result on less gray-level between object and background of infrared image.