In VIVO Fish Diet Discrimination using Selected Hyperspectral Image Classification Methods

摘要

The main aim of this study was to evaluate the performance of different supervised classification methods to discriminate live fish based on their diet received during cultivation using hyperspectral imagery system. 160 rainbow trout were fed either a commercial based diet or completely plant-based diet. Hyperspectral images of the live fish acquired in the spectral region of 394-1090 nm. Spectra were extracted from the region of interest and pre-processed using Savitzky-Golay smoothing algorithm to remove noise. Afterward, three classifiers including support vector machine, random forest and k-nearest neighbors were used. According to the criteria of correct classification rate and kappa coefficient, the support vector machine with linear kernel was achieved the best performance for classirying live fish due to their diet.