SRPAIS: Spectral Matching Algorithm Based on Raman Peak Alignment and Intensity Selection

• Published in: Artificial Intelligence and Security Vol. 13339; pp. 386 - 399
• Main Authors: Sun, Yundong, Tian, Yuchen, Li, Xiaofang, Qu, Rongning, Cheng, Lang, Peng, Shitao, Jia, Jianna, Zhu, Dongjie, Tian, Zhaoshuo
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2022; Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: Big data; Data analysis; Liquor authenticity; Matching algorithm; Micro-spectrometer; Spectral detection
• ISBN: 3031067878; 9783031067877
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-031-06788-4_33

Currently, the frequent occurrence of safety incidents involving food counterfeiting has greatly disrupted the normal market order, infringed on the rights and interests of regular manufacturers and consumers, and even caused personal injury to consumers. Spectroscopic technology can achieve non-contact and non-damaging rapid detection, therefore, leveraging portable spectral matching technology to conduct food detection and analysis has become a research hotspot. Aiming at the problem of unstable matching results caused by instrument laser intensity and control errors in actual spectrum matching scenarios, this paper innovatively proposes a Spectral matching algorithm based on Raman Peak Alignment and Intensity Selection (SRPAIS). First, we innovatively propose a spectral curve pre-processing algorithm based on Raman peak alignment. Before matching, the tested and the target curves are numerically aligned according to the Raman peak, which can greatly alleviate the error of laser intensity caused by instruments and the control systems. Secondly, we innovatively propose a fast-matching algorithm based on an intensity selection strategy, which can further improve the speed and accuracy of spectral matching in big data scenarios. Finally, in the actual liquor-detection scenario, we validated our proposed algorithm through extensive experiments. Experimental results show that our proposed algorithm can significantly improve the accuracy of matching compared with the matching algorithm based on Pearson correlation coefficient, with better discrimination between different samples, and greatly improved stability.