【摘要】 针对高分辨率遥感图像的特点,提出了一种多特征分组融合的可识别性分析方法。该方法首先提取图像的各类特征,包括颜色特征、纹理特征以及形状特征等,然后按照特征分组优化的组合方式分别采用Adaboost进行训练学习,最后将各组特征分类器处理的结果用决策树的方法进行融合获得最终的结果。实验分析对比了单个特征、分组特征以及分组特征融合三种方法的识别率,与前两种方法相比,多特征分组融合的方法具有更高的精度,因而得到结论,多特征分组融合是一种有效的高分辨率遥感图像可识别性分析方法。更多还原

【Abstract】 In view of the characteristics of high resolution remote sensing images,a method of identifiability analysis for multi-feature grouping fusion is proposed. Firstly,this method extracts all kinds of image features,including color feature,texture feature and shape feature,and then,according to the combination way of feature grouping optimization,Adaboost is used to train.Finally,the final results are obtained by using the decision tree to make fusion with the result of each feature classifier. This experiment analyses and compares the recognition rates of three methods,including single feature,packet feature and feature group fusion. Compared with the first two methods,the multi-feature group fusion has higher accuracy. Therefore,it is concluded that multi-feature group fusion is an effective for high resolution remote sensing images.更多还原