An Image Segmentation Algorithm Research Based on Optimized PCNN

To solve the problems of setting complicated parameters and manually adjusting iteration number in the standard pulse coupled neural network (PCNN) model, a revised method named Optimized-PCNN(OPCNN) is proposed in this paper. Firstly, particle swarm optimization (PSO) algorithm is applied to determine the optimal PCNN's parameters for certain image. Then, Otsu algorithm is applied to determine the iterations of PCNN model. In this way, the new method achieves automatic setting of PCNN model and saves a lot of time for scientists who adjust parameters and iteration number manually. This paper uses Otsu method and the origin PCNN method to make comparison with the OPCNN method. After testing on natural light images, microimages and license plate images, Shannon Entropy is used as indicators in quantitative analysis to evaluate the performance of three algorithms. The experimental results show that OPCNN method can retain more details than Otsu method and original PCNN method, and obtain the largest values of Shannon Entropy.