一种高光谱图像压缩投影改进方案  

An Improved Compressive-projection Method of Hyperspectral Images

作　　者：贾应彪[1] 

机构地区：[1]韶关学院信息科学与工程学院,广东韶关512005

出　　处：《韶关学院学报》2016年第8期12-16,共5页Journal of Shaoguan University

基　　金：广东省自然科学基金资助项目(2016A030307044);韶关市科技项目(441-99000311);韶关学院科研项目(314-140685)

摘　　要：在高光谱数据处理中,基于随机投影的降维算法研究开始受到关注,压缩投影主成分分析(CPPCA)是一种由随机投影值重构高光谱图像的有效方法.根据高光谱图像具备的空间相关性,基于CPPCA提出一种新的改进方案.先在空间维把高光谱数据转换至小波域并依据其高低频情况对数据进行分类,再在光谱维选择不同的抽样率参数进行随机投影;重构时,利用CPPCA重构方法分别恢复各类小波域数据,再在空间维进行小波反变换获得高光谱图像.仿真结果表明,与原有CPPCA方法相比,高光谱图像重构质量得到提高,尤其是在抽样率低于0.2的情况下,SNR指标提高超过10 d B.There is increasing interest in dimensionality reduction through random projections for Hyperspectral Images（HSI）. Compressive-Projection Principal Component Analysis（CPPCA） is an efficient receiver-side reconstruction technique that recovers HSI data from encore-side random projections. According to the spatial correlation of HSI,an improved CPPCA method based on single layer wavelet transform is proposed in this paper.The wavelet coefficients of HSI is divided into several subsets and the random projection of different subset is implemented with varying dimensionality,such as a strong degree of dataset reduction for high-pass coefficients subset and a slight degree of dataset reduction for low-pass subset. For the reconstruction,CPPCA is used for each subset and then the HSI could be reconstructed by the inverse wavelet transform. Experimental results with HSI datasets reveal that the proposed method is superior in performance compared to traditional CPPCA.

关 键 词：高光谱图像 随机投影 压缩投影主成分分析 小波变换 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]