贝叶斯理论在多波段SAR图像分类融合中的应用  

Application of Bayesian theory in multiband SAR images fusion for classification

作　　者：岳晋[1] 杨汝良[1] 宦若虹[1] 

机构地区：[1]中国科学院电子学研究所

出　　处：《中国科学院研究生院学报》2008年第2期257-263,共7页Journal of the Graduate School of the Chinese Academy of Sciences

摘　　要：将贝叶斯理论用于多波段SAR图像的分类.分析了常见的乘积方法、平均方法以及中值方法,并在贝叶斯平均方法的基础上,利用SAR图像分类精度与距离因子之间的关系,提出3种改进方法.实验结果表明,多波段融合可以结合各波段的优势和互补信息,获得单波段分类无法获取的分类结果.改进方法通过加权减小了错误分类信息的影响,进一步提高分类精度.In this paper,Bayesian theory is applied to classification of multiband SAR images.The familiar fusion methods like product,averaging and median rules are discussed,and on the base of Bayesian averaging method,three modified rules are proposed according to the relationship between classification accuracy and distance factor of SAR image.Experimental results show that multiband fusion can combine the preponderance and complementary information of each band and achieve classification results which couldn＇t be got in single band.The modified methods decrease the effect of the wrong classification information through weighting and accordingly improve the classification accuracy.

关 键 词：SAR图像 贝叶斯理论 多波段融合 分类 

分 类 号：TN957.52[电子电信—信号与信息处理]