论文部分内容阅读
三重Markov随机场(TMF)模型非常适合处理非平稳、非高斯图像的分割问题.为了降低模型和算法的复杂性,以满足对实测SAR图像处理的实时、稳健和高效的需求,文中提出了一种快速TMF的无监督SAR图像多类分割算法.该算法首先针对SAR图像的乘性斑点噪声,研究了SAR图像四叉树分解的数字特征、阈值选取及分解规则,使得在图像平滑区进行粗分解,而在图像边缘区进行细分解,将图像快速映射成一种新的基于边缘信息的pixon描述,然后再将TMF算法进行扩展,导出了基于边缘信息pixon描述的TMF新的势能函数,最后完成Bayes最大后验模型(MPM)分割.测试数据和实测SAR图像的仿真实验验证了快速TMF算法的有效性.
The Triple Markov Random Field (TMF) model is very suitable for the segmentation of non-stationary and non-Gaussian images.In order to reduce the complexity of the model and algorithm to meet the real-time, robust and efficient requirements of the actual SAR image processing, This paper first studies the digital features, threshold selection and decomposition rules of quadtree decomposition of SAR images based on the multiplicative speckle noise of SAR images. The algorithm makes the rough And decompose it in the edge region of the image, the image is quickly mapped into a new pixon description based on the edge information, and then the TMF algorithm is extended to derive a new potential function of TMF based on the edge information pixon and finally completed Bayes maximum a posteriori model (MPM) segmentation.The simulation of test data and measured SAR images validates the effectiveness of fast TMF algorithm.