Application of image texture for the sorting of tea categories using multi-spectral imaging technique and support vector machine

Abstract

Multi-spectral imaging technique was applied to sorting the green tea category. 320 images were captured at three wavelengths (580, 680 and 800nm) using a multi-spectral digital camera. Entropy values of images were obtained as image texture features. The correction answer rate of least squares-support vector machine (LS-SVM) with radial basis function kernel was up to 100% which was better than those of LS-SVM with linear kernel, partial least squares and radial basis function neural networks, respectively. Results of generation ability test shows that LS-SVM with radial basis function kernel could be effectively used for the application on a few samples. It could be concluded that it is possible to take multi-spectral images of tea and tell which category it is. The whole process is simple, fast, non-destructive and easy to operate.