基于粒子群算法优化PCNN的织物疵点分割

Fabric Defect Segmentation Based on Particle Swarm Optimization Optimized PCNN

探讨基于粒子群算法优化PCNN参数的织物疵点分割。将PCNN中3个参数当作粒子群中粒子,根据经PCNN分割后图像的熵作为PSO的适应度函数,寻找PCNN模型中参数的最优值。分割对比实验从主观和客观两方面验证了该方法的可行性和有效性,并与传统PCNN和OTSU分割方法进行比较。认为:此种方法的分割效果良好,可以有效提高模型的自动化程度。

Fabric defect segmentation based on particle swarm optimization optimized PCNN parameter was discussed. Three parameters of PCNN were used as the particles of particle swarm. The entropy of image after segmenting by PCNN was used as the fitness function of PSO. According to the fitness function of PSO,the opti- mal value of parameter in PCNN model was found. The segmentation contrast experiment has verified the feasi- bility and effectiveness of the method from subjective and objective perspectives. The method is compared with traditional PCNN and OTSU segmentation method. It is considered that the method has better segmentation effect and can improve the automation degree of model efficiently.