基于K-S检测的SAR图像灰度压缩性能评估  

Performance Evaluation for Gray-scale Compression of SAR Images Based on K-S Test

作　　者：王晓军[1] 孙洪[2] 董鹏曙[1] 

机构地区：[1]空军雷达学院预警探测装备系,武汉430019 [2]武汉大学电子信息学院,武汉430079

出　　处：《现代雷达》2006年第7期45-48,共4页Modern Radar

基　　金：国家自然科学基金资助项目(60372057)

摘　　要：根据灰度压缩前后图像直方图间的形态差异,提出一种衡量各种合成孔径雷达(SAR)图像灰度级压缩算法性能的两步检验方法。首先,基于柯尔莫哥洛夫-斯米尔诺夫(K-S)检测原理,在给定置信界限下,选出压缩后图像分布与原始分布满足一致性的算法。其次提出等效均值偏量、等效方差偏量、等效偏态差三个检验指标,对经第一步选择出的方法做进一步定量评价,从而得到最终的算法性能检验结果。计算机仿真表明,该方法能有效衡量压缩前后图像灰度分布函数间的形态差异。Based on the requirement that gray-scale compressed Synthetic Aperture Radar （SAR） image should preserve the distribution of original data, this paper evaluates the performance of gray-scale compression algorithms with a two-step method. Firstly, based on the principle of Kolmogoroff-Smirnoff （K-S） test, under the gamma distribution hypothesis, for a given confidence limits, those algorithms that preserve the original distribution are selected. Secondly three quantitive criteria： equivalent mean error, equivalent variance error and equivalent skewness error to measure the first to the third moment of two histograms and obtain the final results of chosen algorithms are proposed. Experimental result of real SAR image shows the proposed method can successfully measure the distribution preservation capability of tested algorithms.

关 键 词：SAR图像 灰度级压缩 性能检验 K-S检测 

分 类 号：TN957.52[电子电信—信号与信息处理]