Research On Plant Leaves Recognition Based On Color And Texture Features

Plant leaves recognition has the vital significance to study the relationship between plants and illuminate the law of the plant system evolution. At the same time, it can also play a certain role in promoting the development of disciplines related botany and the popularization of plant knowledge. In recent years, the research on leaves recognition in plants constantly made progress at home and abroad, including the choice of leaf images characteristics, the algorithm performance and the design of classifiers. However, the accurate rate of the automatic recognition of plant leaves has still space to improve. This paper will study on the recognition algorithm based on texture and color features of plant leaves.This paper firstly describes the background and significance of plant leaves recognition. And the research progress at home and abroad is reviewed in the field of plant leaves image recognition. Then related technologies about the plant leaves recognition are briefly introduced. For example, the image preprocessing technology of plant leaves and so on.In order to improve the accuracy of plant leaves images recognition, this paper proposes a texture feature extraction algorithm based on the gray-direction co-occurrence matrix and improve a color feature extraction algorithm, then a classifier model based on K-means and SVM is proposed. Finally, the texture and color feature extraction algorithm and the classifier model based on K-means and SVM construct a recognition algorithm of plant leaves. The gray-direction co-occurrence matrix is constructed mainly based on the gray and the changes of angle that leaves texture show when the texture extend from the spindle of leaves to both sides. And the Energy, the Moment of Inertia, the Inverse Difference Moment, the Entropy of Mixing are used to describe the texture features. Improved color feature extraction method is mainly based on the leaves color feature extraction that is based on the Moments about Mean of the HSI color space. It extracts the color feature according to the different distribution of the color in leaf space. The main idea is as follows. Firstly, the leaf is divided into many small regions; Then, the hue and the saturation are chose and combine with spatial position information to extract the color features. Finally, this paper demonstrate the effectiveness of the proposed algorithm by experiment. In addition, this paper also design and implement a plant leaf recognition system based on the leaf recognition algorithm that is proposed in this paper.