基于多分辨率分析提取遥感光谱特征

Feature Extraction of Remote Target Spectra Based on Multi-resolution Analysis

利用子波变换的多分辨率分析方法，将遥感光谱分解为反映目标结构概貌的模糊信号和刻划目标结构细节的锐化信号。结果表明：对目标特征的分析，小波变换方法比FFT更为敏感。

In this paper we describe the spectral feature extraction using the principle of multi-resolution analysis method based on wavelet transform We decompose a set of original spectral data into the blurring signal part that, concentrates most energy and the sharpening signal part represents the changes in the sources signal. The results show that the method can extract target feature with greater sensitivity than the FFT.