Study on Identification of Multiple Pesticide Residues in Lettuce Leaves Based on Hyperspectral Technology

• Published in: Advances in Artificial Intelligence and Security Vol. 1424; pp. 537 - 550
• Main Authors: Cong, Sunli, Liu, Chen, Zhu, Zhi, Hu, Aiyun
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2021; Springer International Publishing
• Series: Communications in Computer and Information Science
• Subjects: Hyperspectra; Lettuce leaves; Multiple pesticide residues
• ISBN: 303078620X; 9783030786205
• ISSN: 1865-0929; 1865-0937
• DOI: 10.1007/978-3-030-78621-2_45

With the long-term irrational use of pesticides, the resistance of diseases and insect pests to pesticides is increasing. The effect of relying solely on a single species of pesticide to control diseases and insect pests is no longer significant, which causes the phenomenon of mixed use of pesticides is becoming more and more common. To solve the problem that current nondestructive methods for detecting a single pesticide residue cannot meet simultaneous multiple pesticide residues, one method based on hyperspectral imaging technology for identifying multiple pesticide residues in lettuce leaves was investigated. In this paper, nondestructive and fast identification for multiple pesticide residues was performed from the angle of spectral analysis. Comprehensively considering the running time, the detection accuracy, the convergence iteration number and the particle number (N) of GSA algorithm, the support vector machine optimized by Gravitational search algorithm (GSA-SVM) model (N = 40) achieved the best performance, with the accuracies of 100% and 96.08% for training set and test set, respectively. The hyperspectral imaging technology combined with GSA-SVM model is feasible for identifying multiple pesticide residues in lettuce leaves, and so hopefully to provide a methodological basis for detecting multiple pesticide residues in other vegetables.