TM遥感图像FLAASH大气校正异常值的改正  

Improvement of outlier in TM image after FLAASH atmospheric correction

作　　者：温素馨 韦玉春[1] 汪美会 

机构地区：[1]南京师范大学地理科学学院,南京210023

出　　处：《测绘科学》2017年第7期165-171,177,共8页Science of Surveying and Mapping

基　　金：国家自然科学基金项目(41471283)

摘　　要：针对经过ENVI的FLAASH模型大气校正后的反射率遥感图像中,经常存在异常值,即负值和大于100的高值,在水体分布众多的图像中尤其明显的问题,以Landsat的TM图像的校正结果为例,设计了改正算法,即对于异常高值采用阈值法进行改正,对于异常负值采用窗口搜索最小正值法进行改正。使用统计方法和NDVI植被指数对改正前后图像进行了对比。与改正前的图像相比,改正后图像进行计算的结果合理,表明算法是可行的和有效的。所提出的改正算法能够行之有效地改正图像中的异常值,为之后的遥感信息提取提供了良好的数据基础,具有一定的研究意义和应用价值。According to the fact that the reflectance after atmospheric correction of FLAASH model in ENVI often exists outliers which are presented by negative and high value greater than 100, especially in the image with large water body, this paper takes the correction image of Landsat TM as an example and presents an image im- provement algorithm. For the abnormal high value, using the threshold value method. For the abnormal negative, using the minimum positive in window around the abnormal pixel as new value. Statistical comparison and contrast of vegetation index NDVI show that the outliers exist in the atmospheric correction images have been improved to the normal range, and the NDVI value is also within the reasonable range and its histogram is more intui- tive. Compared with the image without improvement, the new image turns to be more reasonable, showing that the algorithm is feasible and effective. The proposed improvement algorithm can effectively correct outliers that can provide an efficacious data base for the further study of remote sensing information extraction, and have a certain research significance and application value.

关 键 词：大气校正 FLAASH 最小正值滤波 NDVI 

分 类 号：P237[天文地球—摄影测量与遥感]