Reseaech Of Tree Species Identification Based On The Theory Of Wavelet Transform And SVM

Trees have the important role of purifying air and protecting the environment.At present,the trees are severely damaged and the species of trees are decreasing.Plant classification is very important for the protection of tree species and the conservation of plant diversity.There are some problems in the classification of tree species,such as single tree species identification and low recognition efficiency,this paper proposes a method based on Haar wavelet and Support Vector Machine to identify tree species.In the first stage,the samples of the typical tree species in the north and south regions were collected and the basic dust removal treatment was carried out.Then,through the selfmade image acquisition system,the scanner was scanned and Natural light under the camera to capture the three ways to obtain the leaves of the image,the three ways to obtain the images were named as transmission images,scanned images and natural images.The leaves of the acquired images are removed,remove the noise petiole and denoised,then the axial images of leaves ratio,shape parameter,the complexity of geometric feature extraction,and the use of Haar wavelet to extract the energy,entropy,mean and texture feature.On the basis of feature extraction,construct SVM and BP neural network two models,the final identification results show that the Support Vector Machine in species identification of a BP neural network model in recognition of faster,more accurate identification results,meet the requirements for the efficiency of species identification,identification and classification method in this paper put forward the feasible.