Research Of Classification Of Tea Quality By Computer Vision

The quality of tea mainly depends on the identification of sensory testing, which would have led to the differences in the tea quality by human factors. On the basis of the current research, there is to choose two representative types of tea to quality classification as a study in computer vision. Established the computer vision system, and discussed the image filter, the image of two values and the edge testing method. Abstracted the tea characteristic parameter on shape and color, established a pattern of tea quality classification by ANN, and developed a classification system of tea quality. The main research content as follows:(1)According to computer vision testing samples of tea, established the computer vision system to obtain and analysis tea sample images. It gets the images of dry tea, wet tea and tea soup, which can more precise to the analysis results of tea quality.(2)According tea quality characteristics to establish a set of pattern to identify tea quality the computer vision. It included the shape and color characteristics to dry tea,wet tea and tea soup,especially included the chemical characteristics. Experiment results prove these parameters can effectively identify tea quality;(3)After studying tea quality characteristics,the tea quality accreditation algorithm has been put forward on the basis of BP networks. The experiment of identify tea quality based on BP networks showed good results. Its classification accuracy rate is 95%;(4)To develop a classification system of quality tea by VC++6.0. This software has friendly interface and can operate conveniently. It included the tea image processing, feature extraction and classification of tea.(5)Applied a new system module structure,which managed a project by catalogue and uniformed with background database. It not only operates more conveniently,but also easier to data management.Research shows that the system realizes tea classification rapidly and accurately by computer vision. It is valuable for the researches of other new technology in the tea quality classification.