The research of license plate character recognition based on BP network trained by chaos particle swarm optimization

This paper introduces the current research actuality. To the question of traditional BP neural network's low convergence rate and easily get into minimum, low recognition rate, low efficiency and such problems, so we bring the chaos thoughts to the particle swarm optimization algorithm. Considering the chaos researching with strong local search ability and strong ability of meticulous search, we use chaos particle swarm optimization train the BP neural network and make the BP algorithm jump out of local optimal, improving the BP network's convergence rate and recognition rate. Through setting the appropriate BP network's neuron nodes number, it make the license plate recognition best achieved. Through MATLAB simulation experiment and the results show that this method compared to other license plate recognition methods is better and has a wide application prospect.