Plant Leaves Recognition Method Based On Dimension Reduction Local Binary Pattern And Shape Features Of Leaves

The plant recognition is an important research direction in current plant information science,which is of great significance for the protection of plants.It is almost impossible to use artificial recognition of the large-scale plant image data,and the use of computers for assisted plant identification to improve recognition efficiency is the current research hotspot.Leaves,stems and flowers can extract feature recognition information in the process of plant recognition.Among them,the plant leaf image is a two-dimensional shape and it is easy to store,so it is used in practice.Researchers usually extract the texture features and shape features of leaves images for recognition.In reality,due to the variety of plants and shape similarities,the recognition effect is not satisfactory.In order to resolve the problem that the shape similarities and rotation of plants leaves will lower the accuracy of plant recognition,a method of recognizing plants leaves is proposed,which is based on the dimension reduction LBP algorithm and the shape features of leaves.Firstly,the pre-processes of the collected plant leaf images such as graying,size normalization,segmentation,and contour extraction are performed.The image of the original collected plant leaves are colorful images,which have a large amount of data and are easily affected by the environment and not conducive to computer classification and recognition.In order to obtain the more suitable feature extraction of leaves images,the pre-processing step is very necessary.Then,the LBP algorithm is used to extract high dimensional texture features of leaves.The LBP features are commonly used and effective texture features,which are widely used.At the same time,the original LBP features have high dimensionality and sparseness.The PCA is an efficient dimensionality reduction method.By using PCA to reduce the dimension of LBP,a more efficient dimension reduction LBP feature vector can be obtained.On the basis of that,considering the shape features of leaves,it is widely used in plant classification and recognition research,which is the earliest feature vector for the classification criteria and has been widely used in the research of plant classification and identification.In the face of scale and rotation change,it has invariability.And effectively combining LBP features with shape features of the leaves,it can effectively improve the leaves recognition rate.In the experimental part,it finds the best feature block in the experimental environment of this paper by comparing the experimental results of different feature blocks of LBP.By comparing the experimental results of different feature methods,it is verified that the feature method of this paper has a better recognition effect.And a better classifier is found by comparing the experimental results of different classifiers.In addition,a plant leaf recognition system based on the GUI tool in MATLAB software is made to complete the recognition of a single leaf and batch leaves.