利用水深不确定度探测测深异常值的方法  

The Approach on Detecting Outliers of Multi-Beam Data by Uncertainty

作　　者：黄贤源[1,2] 翟国君[2] 黄谟涛[2] 隋立芬[1] 柴洪洲[1,2] 陆秀平[2] 

机构地区：[1]信息工程大学测绘学院,河南郑州450052 [2]天津海洋测绘研究所,天津300061

出　　处：《测绘科学技术学报》2011年第1期70-74,78,共6页Journal of Geomatics Science and Technology

基　　金：国家863计划资助项目(2007AA12Z326);国家自然科学基金资助项目(40974010)

摘　　要：利用最小二乘支持向量机(LS-SVM)合理地构造出海底趋势面关键在于训练样本的选取。在构造海底趋势面的过程中,提出并实现了一种基于不确定度优化训练样本的方法。为了检验该方法的有效性,选取实测的多波束测深数据进行验证,并与趋势面滤波法进行比较。结果表明,该方法能有效地抑制较大偏差训练样本的影响,构造的海底趋势面更为合理,测深异常值的剔除也更为有效。The seafloor surface could be constructed effectively by LS-SVM depend on the train samples chosen reasonably.In the process of seafloor surface conformation,a new method of optimize samples by uncertainty was presented.Some practical multi-beam data was chosen to verify the correctness and rationality of the new method.The results between trend surface filter and the new method showed that the effective train samples could be chosen and the influence of the sample-outliers could be restrained.The reasonable seafloor surface could be constructed by LS-SVM arithmetic,and then the outliers of multi-beam data could be eliminated effectively.

关 键 词：最小二乘支持向量机(LS-SVM) 趋势面滤波 不确定度 测深异常值 训练样本 

分 类 号：P229[天文地球—大地测量学与测量工程]