基于C-V模型的红外图像自动分割方法研究

为解决红外图像分割中背景噪声及边界轮廓的影响，引入了基于曲线演化理论、水平集方法和M-S分割函数的C-V模型。通过将图像表达为分段常量函数来建立适当的能量函数模型，引入水平集的表示方法，在整个图像域中依据最小化分割寻找全局极小值，可令活动轮廓最终到达目标边缘。由MATLAB实现的仿真结果表明采用C-V模型对红外图像进行自动分割不受边界轮廓线连续性限制，对初始轮廓线位置不敏感，对图像噪声具有很强的鲁棒性，对均匀灰度目标分割效果良好。

Automatic segmentation method of infrared images based on C-V model

To overcome the influence of image noise and edge contour in infrared image segmentation,we introduce the C-V model based on techniques of curve evolution,M-S function for segmentation and level sets method in this paper.We establish an energy model by assuming the image consists of two regions of approximatively piecewise-constant intensities under the level sets framework.By minimizing the energy model to find a global minimum in the whole image region,the active contour has been to evolve to the object boundaries.The emulation results established with MATLAB show that automatic segmentation of infrared images based on C-V model is irrespective of edge continuous information,less sensitive to the location of initial contours and is robust in image segmentation with noise.Good effect has been achieved on image segmentation with statistically homogeneous regions.