矢量量化高光谱数据压缩性能评估

性能评估是高光谱数据有损压缩研究的一个关键问题。本文在分析三种典型的基于矢量量化压缩方案的基础上，以K-means聚类准确率的仿真统计比较了三种方案的性能优劣；提出一种失真标准抽取的性能评估框架，在缺乏背景资料的情况下，该框架可以对压缩方案性能给出直观评价，方便了压缩方案的选择及应用。

Evaluation on the Compression Performance of Hyperspectral Data Using Vector Quantization

Performance evaluation plays an important role in the loss compression of hyperspectral data. Three compression methods using vector quantization are analyzed. Their performances are compared based on the simulation of K-means classification accuracy. Then an evaluation scheme called distortion criteria extraction is proposed. In the condition that no background information could be achieved, the scheme scores the distorted data conveniently. And the scheme can afford facilities for the selection and application of the loss compression methods.