基于子波变换的光谱信息数据压缩方法

DATA COMPRESSION METHOD BASED ON WAVELET TRANSFORM FOR SPECTRAL INFORMATION

本文介绍一种基于子波变换的光谱信息数据压缩方法。利用子波变换的多尺度分析原理，将原始光谱数据分解成集中源信号绝大部分能量的模糊信号和反映源信号变化特性的锐化信号。由于锐化信号只在源信号变化梯度大的区域系数值才较大，其他区域都接近零，只需保存少量的系数，就可以实现数据压缩。用本文方法，对２１种典型地物光谱数据进行了数据压缩实验，在１．０～１．７均方根误差情况下，若压缩结果不编码，压缩比一般为４：１～５：１，编码后可达８：１～１０：１。

In this paper, we describe a data compression method based on wavelet transform for spectral information. Using the principle of multi-resolution analysis, we decompose a set of original spectral data into the blurring signal part and the sharpening signal part. The blurring signal concentrates most energy of the source signal, while the sharpening signal represents the changes in the source signal. As most coefficients of the sharpening signal are close to zero, except very few ones according to the region in which the original signal changes greatly, only very few large value coefficients are needed to store, and a favourable data compression can be achieved. The experiment with 21 typical earth resource spectral data show that a compression ratio between 4: 1 -5: 1 can be gained if the compression results are not coded under the mean-square-root error 1. 0-1. 7 5 a compression ratio between 8: 1--10: 1 can be gained if the compression results are coded.