Tea Leaves Classification Based on Texture Analysis

摘要

An SVM with texture analysis-based feature extraction classification method is presented for identification of fresh tea leaves in this paper. This method is proved to be very efficient and effective in the identification of fresh tea leaves through real experiments. First, the texture characteristic parameters of tea leave images are obtained by texture feature extraction. After that, different categories of fresh tea leaves are identified through SVM training. These texture parameters for texture classification include energy, correlation, and contrast obtained from gray-level co-occurrence matrix (GLCM). Experimental results show that the use of SVM for classification of tea leaves can achieve very good results, and the successful classification rate can be as high as 83 %.