摘要
高光谱图像在取得较高光谱分辨率的同时带来了海量数据,使其压缩成为必需.矢量量化技术在高光谱图像压缩中取得了良好效果,但有计算复杂度高的缺点.针对高光谱图像谱带间高度冗余的情况,本文提出基于特征选择的快速矢量量化算法.该算法在减少运算量同时,能取得和LBG算法相近的压缩效果.实验表明在信噪比略微下降的情况下,计算时间下降了94.32%.
Hyper Spectral Image （HSI） can get fine spectral resolution as well as brings enormous data volume, so compression is necessary. Vector Quantization （VQ） can get good effect in HSI compression, but it has the shortcoming of computing expensively. By using of the fact that HSI has high redundancy in its bands, this paper proposed a feature selection based fast VQ algorithm. It has the advantage of being simple, produ- cing a large computation time saving and yielding compression fidelity as good as the LBG algorithm. The experiment results showed that the runtime reduced 94.32% while the SNR slightly reduced by using our method.
出处
《哈尔滨工业大学学报》
EI
CAS
CSCD
北大核心
2007年第11期1748-1750,共3页
Journal of Harbin Institute of Technology
基金
国家自然科学基金资助项目(60472048)
关键词
高光谱图像
图像压缩
矢量量化
特征选择
hypersepctral image
image compression
vector quantization
feature selection
