摘要
提出一种基于活动轮廓的彩色图像全局分割方法。在已有的向量值图像分割模型中,由于考虑的泛函变分形式是非凸的,计算结果往往陷入到局部极小值点而达不到预期的效果。为了克服这一缺点,提出一种新的变分形式,将向量值图像分割和图像去噪统一到具全局极小能量泛函的框架之中。根据变分的对偶公式,新建的模型容易构造且只需较少的计算量。不同于经典的水平集方法,在实际计算过程中无需繁琐的重新距离化水平集过程。通过对人工图像和真实图像数值结果的分析发现,这种方法具有更好的优势。
It presents a color image based on active contour segmentation method in the existing global. For the vector image segmentation model, considering the variational form is not convex, the calculation re- suits often fall into local minimum point and can not reach the anticipated result. In order to overcome this shortcoming, it proposes a new variational form, the vector of image segmentation and image denoising is unified with the global minimum of the energy functional framework. Based on the variation of dual formu- la, this model is easy to construct and needs less computation. Comparing with the classic level set meth- od, the process become without the cumbersome to distance level set process. Based on artificial and real- world image numerical results analysis, the method has better advantages.
出处
《蚌埠学院学报》
2012年第6期1-4,共4页
Journal of Bengbu University
关键词
活动轮廓
全局
向量值图像
图像分割
active contour
global
vector value image
image segmentation