基于特征变换的高光谱图像压缩快速矢量量化算法(英文)  

A fast feature transformation based vector quantization algorithm of hyperspectral image compression

作　　者：陈雨时[1] 王晓飞[1] 张晔[1] 

机构地区：[1]哈尔滨工业大学信息工程系,哈尔滨150001

出　　处：《黑龙江大学自然科学学报》2008年第3期332-336,共5页Journal of Natural Science of Heilongjiang University

基　　金：Supported by the National Natural Science Foundation of China (60472048)

摘　　要：矢量量化能够得倒良好的压缩效果但有着计算复杂度高的缺点。提出了一种基于特征变换的快速矢量量化算法来压缩高光谱图像。该算法利用了高光谱高维空间大都是空的,可以用较低的子空间来表示这一特点。在较低维数的子空间进行矢量量化会大大降低运算的复杂度。所提出的算法在获得和经典的扩展劳埃德算法效果相当的情况下,大大降低了运算时间。VQ based algorithm can get fine result while it has the shortcoming of computing expensive. A fast feature transformation based Vector Quantization （VQ） algorithm of Hyperspectral Image （HSI） compression is proposed. It makes use of the high dimensional space is mostly empty, so it can be represented by a lower dimensional subspace. If we do VQ at the lower dimensional subspace, the computing complexity can be reduced. The proposed method has the advantage of being simple, producing large computation time saving and yielding compression fidelity as good as or even better than the classical method - generalized Lloyd algorithm （GLA）.

关 键 词：高光谱图像 矢量量化 特征变换 快速算法 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]