Plant Leaf Recognition Based on Naive Bayesian Classification and Linear Discriminant Analysis Model

How to accurately classify plant leaves is the main problem facing the development of plant species identification software. This paper is divided into three parts to study the problem of leaf image recognition: the first part is to select the appropriate identification index. Based on the morphological principle, we establish the leaf data quantitative index system from both geometric features and edge contours. The second part is to determine the core indicators from the index system. We establish a two-objective optimization model from the perspective of shape and edge. The naive Bayesian classification method is used to determine the core indicators and the leaf recognition rate of the core indicators is 86.5675%. The third part is the improvement of the model. We combine the high-dimensional index of texture to improve the naive Bayesian classification model, and establish a linear discriminant analysis model. The leaf recognition rates of the two methods are 91.56% and 98.44%, respectively. By plotting the RP curve, the superiority of the improved blade classification method for solving the classification problem of high-dimensional indicators is illustrated.