Leaf Feature Extraction and Classification Based on Combination Algorithm and Probabilistic Neural Network

In order to solve the problem of low precision in plant leaf identification, a method of plant leaf recognition is proposed based on a combination algorithm and probabilistic neural network. Firstly, the features of the leaf shape are quantitatively extracted by the improved corner point detection algorithm SUSAN, Hough transform, and other methods. Then, the improved probabilistic neural network (PNN) model is established to judge the type of leaves, and the leaves are classified again by using the texture data of leaves in parallel series. The experimental results show that the average recognition accuracy is 92.3%. Compared with other recognition techniques, this method improves the accuracy of leaf recognition.