CBERS-CCD数据土地利用/覆盖信息提取最佳波段选择——以贵州喀斯特山区为例

Optimal Band Selection on CBERS-CCD Data in Land Use and Land Cover Information Extraction——A Case of Guizhou Karst Mountainous Area

最佳波段选择是遥感数据处理重要基础工作,不仅影响遥感信息提取目视解译效果,而且影响计算机分类精度和效率。以贵州喀斯特山区CBERS-CCD数据土地利用/覆盖信息提取为例,依据最佳指数(OIF)模型,基于波段间信息量和地物间可分性试验,从定性和定量角度评价试验结果,得出:①CBERS-CCD数据所有可能3个波段组合中,B3-4-1波段组合RGB合成假彩色影像最适宜多数土地利用/覆盖类型信息提取;②CBERS-CCD数据第5波段对水田和草地等地类信息提取有重要贡献,应予充分利用;③不同波段组合对不同地类信息提取贡献有别,仅仅依靠某个波段组合即便是最佳波段组合也难以获得很高精度,需要继续研究CBERS-CCD数据与其它数据融合,以及决策树与多特征提取等多种遥感信息提取方法。

It is an important and basic work of remotely sensed data processing that optimal band selection influences the result of visual interpretation and the accuracy and efficiency of computer classification.The Guizhou karst mountainous area was studied,based on the OIF model,the optimal band selection of CBERS-CCD data in land use and land cover information extraction was obtained,and some useful conclusions were got： firstly,in the all possible three-band combinations,B3-4-1 with RGB is the best one for land use and land cover information extraction;secondly,the 5th band of CBERS-CCD data is very helpful to certain information extraction,such as paddy field and grassland,etc.And thirdly only the optimal band combination to extract information can not get enough classification precision and advanced information fusion and extraction methods to CBERS-CCD data needs to be exploited in the future.