Green plant segmentation in hyperspectral images using SVM and hyper-hue

Abstract

Green plant segmentation plays an import role in hyperspectral-based plant phenotyping, however, this topic is not given enough consideration. Existing image segmentation methods are dependent on data types, plants and backgrounds and might not utilise the power of hyperspectral data. We proposed a one-class support vector machine classifier combined with a pre-processing method named hyper-hue to segment green plant pixels in hyperspectral images. Experimental results showed that his method can segment green plants from backgrounds with fewer errors and therefore could be used as a general method for hyperspectral-based green plant segmentation.