Quality Detection of Beef Based on Spectral Signal Quantile Graph Network Analysis Method

Liu Jie,Xu Huibin,Bao Shengqiao,Wang Hao,Li
DOI: https://doi.org/10.1109/ccdc49329.2020.9163947
2020-01-01
Abstract:Under the viewpoint of the duality of time series and complex networks, this paper proposed a novel method to map spectral signals to complex network maps for the problem that `products doped with viscera in meat products are difficult to detect by conventional food spectrometers. The associated network graph analysis method of the quantile graph extracts the 'intrinsic geometric topological characteristics' of the food spectrum signals obtained by the detection. It conducts the quality inspection research of the remanufactured meat products as the 'sensitive characteristic quantity' of the industrial inspection. Numerical experiments show that, based on the data collected by the conventional spectrometer, the single feature extracted by our algorithm can successfully distinguish the pure beef and the doped beef with different animal internal organs. This method is different from some existed methods reported in the existing quality inspection literatures. It is not necessary to start monitoring the sample as a mature product by using algorithm proposed in our paper, which has the remarkable advantages of being fast, simple, reliable and easy to perform.
What problem does this paper attempt to address?