基于HJ-1A卫星HSI影像的阿克苏地区主栽经济林树种识别研究

Classification of Forest Tree Species in Akesu Region Basin Based on HJ-1A Satellite HSI Image

基于HJ-1A卫星HSI数据,以阿克苏地区主栽3种经济林树种为研究对象,运用最佳指数法(Optimum Index Factor)和波段指数法(Band Index)对HJ-1A卫星的HSI影像进行敏感波段的筛选,并对南疆环塔里木盆地主栽经济林树种进行了识别研究。结果表明,不同波段选择方法筛选出的敏感波段集中在红光和近红外波段范围,且对经济林树种都具有良好的可分性;敏感波段组合(28-40-77、28-54-75、20-40-58)对经济林树种总分类精度均达到了70%以上,基于TM432划分波段的波段指数法筛选的20-40-58组合,总分类精度最高(OA=72.4647%),Kappa系数为0.5838。

Based on HJ-1Asatellite HSI data,three cultivars forest tree species in Aksu region were studied,the optimum index factor（OIF）and band index（BI）of the HSI image HJ-1Asatellite were applied to recognize and research cultivars of the main forest tree species in Tarim Basin in southern Xinjiang.The results showed that the sensitive band screened with different band selection method for screening concentrated in red and near-infrared wavelength range and forest tree species have good separability;Sensitive band combination（28-40-77,28-54-75,20-40-58）of forest tree species overall classification accuracy have reached more than 70%.The combination of 20-40-58 screened by band index number method based on TM432 band have the highest accuracy（OA =72.4647%）in overal classification,the Kappa coefficient was 0.5838.