节点文献
多尺度边缘检测方法提取近红外光谱信息特征的研究
The Study of Feature Extraction of Near-Infrared Spectroscopy with Multi-Scale Edge Extraction Method
【摘要】 利用小波变换中的多尺度边缘检测方法对甲硝唑药片近红外光谱进行了信息特征的提取。通过偏最小二乘法建立所提取的特征与甲硝唑浓度之间的数学模型,并进行精度分析。结果表明,小波变换多尺度边缘检测能够有效分离并提取光谱信息特征,降低噪声,为一种有效的光谱信息特征提取方法,为解决近红外光谱中光谱叠加严重和信息率低等难题提供了一种思路。
【Abstract】 The authors utilized a multi-scale edge extraction method to extract feature in formation of nearinfrared spectroscopy. Applying partial least-squares to chara cteristic data obtained, the authors set up a linear mathematical model of the c orrelation between NIR spectrum and metronidazole contents. The result shows that the multi-scale extraction method can separate and extract feature inform ation, and can be seen as an effective method of spectrum information extraction . The present study offers a new solution to the problems in nearinfrared spectroscopy technolog y.
【Key words】 Feature extraction; Edge extraction; Wavelet Transform; NIR radiation;
- 【文献出处】 光谱学与光谱分析 ,Spectroscopy and Spectral Analysis , 编辑部邮箱 ,2005年06期
- 【分类号】TN911
- 【被引频次】8
- 【下载频次】297