基于局部高斯拟合的主动轮廓模型利用图像的均值和方差信息来拟合图像信息。与只利用图像灰度均值信息建模的主动轮廓模型相比, 该模型能很好地分割复杂的医学图像。但该模型仅利用了图像的局部信息建模, 因此收敛速度比较慢; 并且在建立能量泛函时, 采用传统的Heaviside函数, 分割精度不高。针对这些缺陷, 在改进Heaviside函数的基础上, 引入全局高斯拟合项, 并且对局部高斯拟合项和全局高斯拟合项的权重系数均采用自适应的方法进行调整, 得到基于局部和全局高斯拟合的主动轮廓分割模型。改进模型不仅能有效分割均值相同但方差不同的图像, 还能有效分割质量较差的医学图像, 并通过实验检验了改进模型的性能。

The active contour model based on the local Gaussian fitting utilizes the average and variance information to fit the image information. Compared with the traditional active contour models which only utilize the gray average information, this model can segment the complex medical image successfully. However, this model only utilizes the local information of image to model, so the convergence speed is slow. In addition, the traditional Heaviside function is utilized to establish the energy function, which leads to the limited segmentation accuracy. Aimed at these defects, the global Gaussian fitting term is introduced to improve the Heaviside function. Using the method of adaptive adjustment, an active contour segmentation model based on the local and global Gaussian fitting is obtained. The improved model can not only segment the images with same average but different variance, but also segment the inferior medical images effectively, and the performance of the improved model is verified by experiments.