Study On Varieties Identification Of Ornamental Chrysanthemum Based On Digital Image Processing

Ornamental chrysanthemum is deeply loved by many people for its beautiful color and various shape, however, there are too many ornamental chrysanthemum varieties in the world, just in China there are more than 3000 varieties, so it’s very difficult for people even researchers to identify all kinds of ornamental chrysanthemum. At present, ornamental chrysanthemum is identified mainly by specialists with professional knowledge, but the operation is very complex, and it’s very difficult for people to control so much knowledge. Therefore, the exploration of using computer technology for identifying ornamental chrysanthemum has great significance.A new method of identifying ornamental chrysanthemum by using digital image processing technology was studied in this paper. Firstly, an ornamental chrysanthemum image acquisition device was designed in our paper, which can overcome the influence to image quality from unstable ambient light, and provide a simple background for ornamental chrysanthemum images. Then according to the characteristics of ornamental chrysanthemum, ornamental chrysanthemum’s features were extracted through three aspects: color, shape, and texture. As to the color features, color composition feature can be extracted based on HS color space, and color space feature can be described through flatness; as to the shape features, convex hull was used to describe the shape characteristics of ornamental chrysanthemum; as to the texture features, based on the knowledge of gray level co-occurrence matrix, the related parameters such as energy, entropy, contrast and correlation were selected as texture feature description. And then we got a 14 dimension feature vector through feature fusion. In the next step, KNN classifier and SVM classifier were used to identify 20 ornamental chrysanthemum varieties, and found that both them have high correct recognition rate. At last, the recommended function of ornamental chrysanthemum varieties was added in the process of software design, which improved the final correct recognition rate.