Abstract: Recently, researching for compression Hyper Spectral Image（HSI） technology becomes more and more popular, because HSI needs large storage space. In order to solve the above problem, a new method of compression HSI is proposed, which is based on Discrete Wavelet Transform（DWT） and Tucker decomposition. It makes full use of frequency domain and space domain information of HSI, and wavelet transform parameter of HSI frequency band is decomposed by Tucker decomposition. Firstly, HSI is decomposed into different sub images. Then, sub image is compressed with decomposing Tucker. Finally, effectiveness of the method is evaluated in experiments with real HSI, and compared with other algorithm, effectiveness and performance of the new algorithm is more reliable and better. In addition, new compression HSI method has a good performance in supervised classification.

Key words: compression, Hyper Spectral Image（HSI）, noise reduction, wavelet transform, Tucker decomposition

摘要： 由于超光谱图像（HSI）的大数据量，HSI压缩技术的研究近年来越来越受到关注。鉴于此，提出了一种基于离散小波变换（DWT）和Tucker分解的HSI压缩方法。充分利用HSI频域和空域的信息，对HSI频带的小波变换系数进行Tucker分解，先利用小波变换将HSI分解为不同的子图像，然后利用Tucker分解实现子图像的压缩；最后用实际的HSI对算法的有效性进行评估。与其他算法的比较结果表明该算法具有更好的性能；实验还显示了压缩HSI在监督分类方法中的作用。

关键词: 压缩, 超光谱图像, 噪声抑制, 小波变换, Tucker分解