光谱信息数据压缩方法—二真值线性预测法  

DATA COMPRESSION METHOD FOR SPECTRAL INFORMATION-TWO-TRUE-VALUE LINEAR PREDICTION

作　　者：钱神恩[1] 李树秋[1] 王汝勤[1] 戴逸松[2] 

机构地区：[1]中国科学院长春光学精密机械研究所,长春130022 [2]吉林工业大学

出　　处：《光谱学与光谱分析》1992年第2期9-20,共12页Spectroscopy and Spectral Analysis

摘　　要：本文针对线性预测压缩存在的游长短、压缩比低、恢复精度差等问题,提出了一种新的压缩方法—“二真值线性预测法”,其特点是恢复精度高、压缩比大、算法简捷、实时性好,适用于可由线性逼近表示的测量过程的实时压缩。经对典型光谱数据压缩实验表明,平均游长一般比线性预测压缩高1倍,在压缩恢复过程中可保留光谱特征信息,在1%恢复精度时,压缩后数据未编码的情况下一般可获得大于2:1的压缩比,编码后可获得4:1～5:1的压缩比。In view of the fact that the data compression with linear prediction yields a short run-length, low compression ratio and poor reconstruction precision, a new compression method-two-true-value linear prediction (TTVLP) is advanced in this paper. The features of this method are with large compression ratio, high reconstruction precision, simple and direct algorithm. It can in real-time compress data which can be stated by linear approximation. The experiments with compressing 21 typical earth resource spectral data show that the average run-length with TTVLP is usually one time greater than that with linear prediction and the innate information of spectrum can be preserved in compression/decompression, at least 2:1 compression ratio can be gotten if the compression results arc not coded; 4:1—5:1 compression ratio can be gotten if the compression results are coded.

关 键 词：数据压缩 线性预测 光谱信息 

分 类 号：O433.1[机械工程—光学工程]