基于小波包分析的光谱识别方法研究

光谱信号的自动识别技术是光谱分析技术中的重要组成部分, 是指运用模式识别方法，借助计算机技术，对相同测定条件下的光谱信号进行比较，根据信号之间的相似程度，从而得出两者之间化学组成关系的技术，主要应用于光谱的定性分析，也可用于光谱的验证。文章以符合朗伯-比尔定律的光谱信号为研究对象，简单介绍了光谱识别技术的基本原理和方法，随后为简化识别难度，进行了光谱信号的归一化处理，再在概述小波分析基本原理的基础上，提出了采用小波包分析的技术对光谱信号进行相关特征提取的方法，并根据统计学知识，得出了计算标准特征向量和允许误差向量的公式，然后运用二叉树分级判别的方法，实现了光谱信号的快速识别，最后举例对该方法进行了说明。

Spectral Signal Recognition Based on Wavelet Packet Analysis

Automated recognition of spectral signal is an importance part of spectral analysis, which jnvolves comparing spectral signals under the same conditions and obtaining relation of chemistry constitution according to similarity by pattern recognition method and computer technique, and is useful to determine the nature of spectral signal and validate spectral signal. This paper takes spectral signal according with Lambert-beer' law as object, introduces basic theory and method of recognition of spectral signal in brief, normalizes spectral signal to decrease difficulty, introduces wavelet analysis theory, puts forward the method of wavelet packet analysis to gain features of spectral signal, obtains formula to compute feature vector and error vector by statistics method, then makes use of binary tree' mode distinguishing step by step for quick recognition, and finally gives an example to explain.