摘要： 针对多通道的图像分割提出一种新的活动轮廓线提取算法。对轮廓线长度采用Munford-Shah最小化泛函加上图像每个通道上的拟合误差之和。与传统C-V方法类似，该算法不需要图像的梯度信息就可以检测物体的边缘，解决了传统C-V方法不管从哪个尺度空间都无法完全分割彩色图像中物体的问题。并通过改进初始化函数提高了分割的速度。

关键词: C-V模型, 图像分割, 多通道, 活动轮廓

Abstract: This paper proposes an active contour algorithm for object detection in multi-channel images. The model is an extension of the scalar C-V algorithm to the vector-valued case. The model minimizes a Mumford-Shah functional over the length of the contour, plus the sum of the ?tting error over each component of the Multi-channel image. Like the C-V model, the vector-valued model can detect edges with or without gradient. It solves the problem where the model detects multi-channel objects which are undetectable in any scalar representation. It improves initial function to increase segmentation speed.

Key words: C-V model, image segmentation, multi-channel, active contour

中图分类号: 

• TP391.41