The Study Of Tea-leaf And Tea-stalk Image Recognition And Classification Based On Multi-features And Multi-classifiers

The tea industry is one of the unique industries in China. And the tea culture is popular with people all over the world. In this study, in order to effectively make the tea-leaves and tea-stalks sorting technology of automation, high precision and high efficiency, computer technologies such as image processing and pattern recognition, etc. are employed, the sample image database was established, suitable effective feature vectors such as the color and shape characteristics are selected, and the more simple and effective classifier combination of its experimental verification is used to testify its efficiency. The experimental results show that the tea-leaves and tea-stalks sorting technology is of good recognition effect and low consumption when sorting.First, this paper introduces digital image preprocessing technology for the tea-leaf and tea-stalk, stresses analyzing the foreground segmentation and the binarization method, discusses the main morphological processing method, and discusses the methods of image classification information database established. Combined with the method of image processing technology for the tea-leaf and tea-stalk, this paper puts forward the specific image feature extraction and selection method for the tea-leaf and tea-stalk. Secondly, this paper discusses the characteristics of multiple classifier combination technology. Through the feasibility analysis of the multiple classifier fusion, this paper introduces several kinds of commonly used classification techniques, namely, the least risk bayes classifier, minimum distance classifier and support vector machine (SVM) classifier and so on, at the same time, the current method of multiple classifiers combination and that adopted in this paper are discussed. Finally, combined with the related data from the previous tea-leaf and tea-stalk image pre-processing, the effective leaf stalks characteristics and leaf stalks classifier combination model suitable for this study are employed. Through the comparative experiment on multi-characteristics classifiers and single characteristic classifiers, the paper proves the feasibility of the proposed classification method.The experimental results show that the proposed method to some extent can help to classify the tea-leaf and tea-stalk, reduce the rate of mixing them, improve the sorting efficiency of tea-leaf and tea-stalk and accuracy and it has certain actual application value.