High resolution remote sensing image classification based on mind evolutionary algorithm optimized BP neural network

Abstract

The traditional BP Neural Network has some problems, such as slow learning convergence speed and easy to fall into local extremum, and it needs to be optimized in order to achieve ideal effect in high-resolution remote sensing image classification. Mind Evolutionary Algorithm (MEA) is an intelligent algorithm that retains the idea of population evolution of genetic algorithm and solves the premature problem of genetic algorithm. Combining MEA with BP neural network to construct MEA-BP model, the essence of which is MEA is used to optimize the BP neural network weight threshold, and the convergence and alienation operations are used after the initial population is generated by MEA. The classification experiment of high-resolution remote sensing image is carried out by using MEA-BP model. The results show that the BP neural network optimized by MEA obtains the highest producer accuracy, user accuracy and overall classification accuracy, in which the producer accuracy and user accuracy are stable above 85%, the overall classification accuracy reaches 91.66%, and the kappa coefficient also reaches 0.8970, which reflects a good classification effect.