中文摘要：

提出了一种新的基于多分辨率马尔可夫场（multiresolution Markov random field，MRMRF）模型的小波域多光谱纹理影像分割方法。该方法采用可变的权重参数连接小波分解的多尺度特征场和标记场，通过直接映射，将上一尺度的分割结果作为下一尺度的初始结果，最细尺度上的分割结果作为算法的最终分割结果。实验表明，本文算法能够降低多光谱纹理图像分割的错误率。

英文摘要：

A new method based on multiresolution Markov random field（MRMRF） in wavelet domain for multispectral textured image segmentation is presented, in which a variable weighting parameter is employed to combine the character random field and the label random field. And a segmentation result on a coarser resolution is employed as the initial segmentation of the nearest finer resolution by directly projecting the segmentation at the finest resolution. Experimental results have shown the quality of the proposed method to reduce the segmentation error rate.