A Recognition Model of Red Jujube Disease Severity Based on Improved PSO-BP Neural Network

In order to improve the accuracy of the red jujube disease recognition, the study establishes a recognition model of disease severity, with the improved Particle Swarm Optimization Back Propagation(PSO-BP) neural network combined with color and geometry characteristic parameters of red jujube tree leaf disease spot. Mutation operator and linear decrease inertia weight are combined to improve the performance of PSO, a new improved PSO is formed to get optimal neural network weights and thresholds. The experimental results show that the accuracy and performance of red jujube disease recognition model is improved. The slight, general and serious disease reached separately 87.6%, 82.4% and 94.0%.