Classification of Collection Walnut Based on GLCM and SVM

Abstract

In order to complete the classification and matching of walnut more efficiently, this paper proposes a texture feature extraction and classification method based on gray level co-occurrence matrix and support vector machine (SVM) classifier. The method extracts the characteristic parameters of walnut image, such as energy, contrast, correlation, inverse different moment and so on, and then uses BP neural network and support vector clustering model to train and test the extracted characteristic parameters, in order to check the extraction of matrix parameters. The results of feature extraction and classification test of 300 samples of Mangjian lion head, Guanmao walnut and Jixin walnut by Matlab simulation software show that the recognition rate of feature parameters extracted by gray level co-occurrence matrix can reach 93.3%. The recognition rate of SVM is higher than BP neural network classification, which means SVM clustering method is more suitable for textural classification matching.