基于区域特征的自适应图像分割方法

本文提出一种以图像统计假设模型和系统条件熵相结合的自适应图像分割新方法，该方法能依据图像统计特征自适应地改变自组织特征映射网络的权值，解决卫传统图像分割方法所存在参数需要人工调整、无法获得最优解的问题。实验结果表明，本文方法对于复杂图像的分割具有很好的鲁棒性和有效性。

An Adaptive Approach to Image Segmentation Based on Region Features

A new adaptive image segmentation method is presented in this paper,which is based on the image stochastic hypothesis model in the combination with the system conditional entropy.And the weight of self-organizing feature mapping network can be updated adaptively on the basis of image stochastic characteristics.The problems of interactive parameters and local minimum in conventional methods have been solved. The experimental result shows that the method in this paper possesses robust and efficient for various types of complex images.