Study On Plant Identification Based On Image Features

Plant identification is closely related to people's life.The traditional method of plant identification is complicated and not conducive to popularization.With the rapid development of computer image processing and pattern recognition technology,the automatic recognition of plant species based on image processing becomes possible.In recent years,more and more researchers pay more and more attention to the research of automatic recognition technology based on plant images.Based on this,we study the characteristics of plant based image in this paper.Firstly,we summarize some common features in recent years.Secondly,according to the characteristics of entropy sequence extracted by PCNN model,we make some improvements to the traditional extraction methods,from attenuation,weight matrix expansion and other aspects of the exploration,and several related models of PCNN are compared.Then,we carry out a single feature evaluation experiment according to the characteristics of the paper,so that the researchers can select the recognition features better.Finally,according to the characteristics,mainly combinations of two features and three features were studied,and compared with SVM classifier and KNN classifier in Flavia library and ICL Library of leaves.In the end,we not only prove that some of the improvement of the entropy sequence characteristics of PCNN is effective,but also show that the combination feature can improve the recognition rate.