Extraction model of forest features based on mutation and bidirectional particle swarm optimization

Yan Li¹, Lihai Wang², Yanqiu Xing²

COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(12A) 215-220

¹College of Information and Computer Engineering, Northeast Forestry University, Harbin, 150040, Heilongjiang, China

²Forest Operations and Environmental Research Center, Northeast Forestry University, Harbin, 150040, Heilongjiang, China

Although the existing forest feature extraction and classification model has a certain effect, but it still exists problems such as accuracy is not high, speed is slowly and so on. According to this problem, this paper proposed an extraction model of forest features based on mutation and bidirectional particle swarm optimization.First, we use mutation operators of genetic algorithm and the Sigmoid function of neural network to make a dynamic adjustment, in order to avoid the particles into a precocious state. Then according to this paper the original algorithm of the initial population is improved on the basis of the use of the two-way optimization strategy, and put forward the speed optimization strategy to help it get a local optimal solution timely when it appeared premature phenomenon and use the optimization strategy of particle effect to enhance the convergence accuracy. Finally the gauss perturbation theory is introduced to improve the convergence of the algorithm when the original algorithm falls into local optimum with optimize learning strategy to make it jump out. Through the simulation experiments, it shows that the proposed forest feature extraction model based on the variation feature subset and two-way optimized particle swarm algorithm with higher precision, better convergence performance