Identification of tea white star disease and anthrax based on hyperspectral image information. (3rd November 2020)

• Record Type: Journal Article
• Authors: Lu, Bing; Jun, Sun; Ning, Yang; Xiaohong, Wu; Xin, Zhou
• Is Part Of: Journal of food process engineering. Volume 44:Number 1(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2020-11-03
• Subjects: Food industry and trade -- Periodicals; Food -- Analysis -- Periodicals; 664.005
• DOI: 10.1111/jfpe.13584
• Languages: English
• ISSNs: 0145-8876

Abstract

Abstract: Hyperspectral images were used to identify the two similar diseases of tea white star disease and anthrax in this research. The average spectra of healthy leaves, white star disease, and anthrax leaves were collected, respectively. It was found that the average spectrum of white star disease and anthrax had strong morphological correlation and poor classification results. Then, the mask technology was used to segment the diseased region of leaves in order to get the best region of interest. After that, the average spectral separability of diseased region was significantly improved. Finally, through the comparison of the classification results between support vector machine and extreme learning machine (ELM), it was found that the ELM model based on neural network structure got the best identification results, and its classification accuracy reached 95.77%. This study provides a new method to identify similar diseases of leaf plants. Practical Applications: There is a certain similarity in disease characteristics between white star disease and anthracnose disease of tea. The similarity leads to a low accuracy in the classification and identification of diseases using hyperspectral technology. In order to solve this problem, this research proposed a spectrum extraction method based on the region of interest of the spots region. The experimental results showed that the average spectrum obtained from the leaf spots region could significantly improve the characterization of tea white star disease and anthrax, and the classification accuracy of the prediction model was significantly improved. This study provides a theoretical reference for the identification of tea similar diseases. Abstract :