基于小波多尺度分解的红外光谱谱图比对检索方法研究

谱图比对检索是指计算提问谱和数据库中标准谱图的相似度,以查出与提问谱相同或近似的谱图,其中,提问谱的质量是影响谱图比对检索结果的一个重要因素.从削弱谱图的无关信息对谱图比对检索的影响、提高检索准确性出发,以提问谱噪声为例,分析了谱图噪声对相似度计算结果的影响;针对红外光谱谱图经过小波多尺度分解后,其有用信息主要集中在低频部分,高频部分主要是噪声的特点,提出了基于小波多尺度分解的红外光谱谱图比对检索方法;分析了小波分解层数对相似度计算结果的影响;选择去噪整波形比对检索作为参照方法,利用MATLAB 7.0分别设计了这两种比对检索程序,实验结果证实了基于小波多尺度分解的红外光谱谱图比对检索方法的可行性.

Research on infrared spectrum intercomparison search method based on wavelet multi-scale decomposition

Spectrum intercomparison search means calculating the similarity of question spectrum and standard spectrum in the database to find the same or similar spectrum as the question spectrum. To weaken the impact of irrelevant information on spectrum intercomparison search results and improve the search accuracy, the question spectrum noise is taken for example and the impact of spectrum noise on the calculated similarity results is analyzed; for the characteristics that when infrared spectrum is multi-scale decomposed, the useful information is mainly concentrated in the low-frequency part and the high-frequency part is mainly noise, the infrared spectrum intercomparison search method based on wavelet multi-scale decomposition is proposed; the impact of spectrum decomposition level on the calculated similarity results is analyzed; the entire waveform search method is used as reference method and the search program of both methods is designed with MATLAB 7.0 and the feasibility of infrared spectrum intercomparison search method based on wavelet multi-scale decomposition is confirmed by the experimental results.