基于独立成分分析的高光谱图像数据降维及压缩ICA.pdf

第 29 卷第 12 期 电 子 与 信 息 学 报 Vol.29No.12 2007 年 12 月 Journal of Electronics Information Technology Dec... .2007 基于独立成分分析的高光谱图像数据降维及压缩 冯 燕 何明一 宋江红 魏 江 (西北工业大学电子信息学院陕西省信息获取与处理重点实验室 西安 710072) 摘 要：该文提出了一种以高光谱图像分析为目标的基于独立成分分析的高光谱图像降维和压缩方法。该方法首先 通过独立成分分析提取高光谱数据的光谱特征实现高光谱图像降维，再对降维后的图像采用预测和自适应算术编码 的方法进行压缩。对 220 波段和 64 波段高光谱数据的实验结果表明，该方法与基于主成分分析的降维相比，压缩 比有所提高，特别是更有利于后续的分析处理，但峰值信噪比有所降低。 关键词：高光谱图像压缩；独立成分分析；主成分分析；降维；分类 中图分类号：TP751.1 文献标识码：A 文章编号：1009-5896(2007)12-2871-05 ICA-Based Dimensionality Reduction and Compression of Hyperspectral Images Feng Yan He Ming-yi Song Jiang-hong Wei Jiang (Shaanxi Key Lab of Information Acquisition and Processing, Northwestern Polytechnical University, Xi’an 710072, China) Abstract: This paper proposes a dimensionality reduction and compression method of hyperspectral images based on Independent Component Analysis(ICA) for hyperspectral image analysis. At first hyperspectral features are extracted using ICA and dimensionality reduction is accomplished. Then, dimensionality reduction images are compressed by the predictive code and adaptive arithmetic code. The experimental results by using 220 bands and 64 bands hyperspectral data show that the method achieved higher compression ratio, more strong analysis capability and lower peak signal-to-noise ratio than dimensionality reduction based on Principal Components Analysis(PCA). Key words: Hyperspectral image compression; Independent Component Analysis(ICA); Principal Components Analysis(PCA); Dimensionality reduction; Classification 1 引言