摘要
针对干旱半干旱区水域的重要性,开展基于环境减灾卫星的水环境遥感识别监测已成为目前水资源领域中的重要任务。该文利用LBV变换能显著突出地物信息的这一特征,以环境卫星为数据源探讨LBV变换方法。在分析L、V、B3类分量的基础上,以V-B及L-B为特征空间,结合波段阈值、二维散点分布与回归拟合方法,识别研究区的水域分布信息。结果表明,使用单一B分量阈值法能较完整地提取出水域信息。利用水体集群聚集程度高的方式提取的水体信息,在V-B及L-B特征拟合下也能完整地提取研究区内的水域信息。比较3种水体信息提取方法,无论是在误提取率还是在水体提取精度上,L-B特征拟合方法提取的效果最好,其次是V-B特征关系,B分量阈值法提取精度最低。
Because of the importance of water in the arid and semiarid regions, to carry out water environment remote sensing monitoring based on environmental mitigation satellite (HJ) has become an important task in the field of water resources. Using the feature that LBV transform can get prominent feature information, this paper explored LBV transformation method. Combining with single-band threshold, two-dimensional feature space and regression, the area of water body information was identified on the basis of analyzing each component. The results showed that using single threshold of B component could extract the water information completely. With high degree of water cluster utilization, the information extraction could be completed in the water area under the condition of V-B and L-B feature fitting. Through comparison of three water extraction methods, it was found that the result of L-B feature was the best, V-B feature was the second, the extraction accuracy of B component threshold was the lowest.
出处
《测绘科学》
CSCD
北大核心
2016年第10期165-169,共5页
Science of Surveying and Mapping
基金
新疆基础测绘工程塔里木河流域地表覆盖变化监测试点项目
关键词
LBV变换
特征空间
回归拟合
水体信息
LBV transformation
feature space
regression
water extraction