Nondestructive detection of mango soluble solid content in hyperspectral imaging based on multi-combinatorial feature wavelength selection

• DOI: 10.1556/066.2023.00014
• Publication Date: 2023-08-09T11:12:49Z
• Authors: J.J. Lin; Q.H. Meng; Z.F. Wu; S.Y. Pei; P. Tian; X. Huang; Z.Q. Qiu; H.J. Chang; C.Y. Ni; Y.Q. Huang; Y. Li
• Keywords: Variable elimination; Content (measure theory); Feature (linguistics)

Abstract This paper explores the prediction of soluble solid content (SSC) in visible and near-infrared (400–1,000 nm) regions Baise mango. Hyperspectral images mangoes with wavelengths 400–1,000 nm were obtained using a hyperspectral imaging system. Multiple scatter correction (MSC) was chosen to remove effect noise on accuracy partial least squares (PLS) regression model. On this basis, characteristic mango SSC selected competitive adaptive reweighted sampling (CARS), genetic algorithm (GA), uninformative variable elimination (UVE), combined CARS + GA-SPA, UVE-SPA, GA UVE-SPA wavelength methods. The results show that MSC-CARS GA-SPA-PLS can reduce redundant information improve computational efficiency, so it is an effective method predict mangoes.