基于SWBCT和投影特征的遥感目标识别

方向信息提取与小样本问题是遥感目标识别与应用的瓶颈,基于平稳小波的Contourlet变换(Stationary Wavelet Based Contourlet Transform,简称SWBCT)与投影特征相结合,本文提出了一种新的遥感目标特征提取与识别方法.首先采用改进的Contourlet变换SWBCT对遥感图像进行分解,然后对分解子带进行方向投影并求投影向量的统计特性作为目标特征,进而利用K近邻分类器进行目标识别.与小波、Contourlet等代表性方法进行比较,实验结果与分析表明,该方法不仅具有较高的识别率,而且在小样本情况下,依然可以得到较好的结果.

REMOTE SENSING TARGET RECOGNITION BASED ON SWBCT AND PROJECTION FEATURE

Directional information extraction and small sample problem are the key problems in remote sensing object recognition and applications.Based on the stationary wavelet based contourlet transform(SWBCT)and projection,a novel remote sensing target recognition method was put forward in this study.Firstly the new method decomposed the remote sensing image by the improved Contourlet transform SWBCT.Then it extracted statistical features of the subbands projectioned in different directions.And K-nearest classifier was used to classify the features.The simulation results show that the new algorithm has higher recognition power compared with the wavelet,Contourlet and some other well-known methods.Meantime,in the condition of small scale samples,our method still has very well results.