Color Characterization of LCDs Based on Improved PSO-Elman
The LCD color characterization can realize the accurate display of the same image on different devices.To solve the problems of complicated model establishment and poor model robustness leading to low characterization accuracy in LCD color characterization,a conversion model from RGB color space to CIEXYZ color space(ACOPSO-Elman)is proposed based on an improved PSO-Elman neural network approach.Firstly,the adaptive adjustment function of inertia weight and learning factor is constructed according to the particle population size and particle posi-tion relationship to improve the global optimization-seeking ability and convergence speed of PSO algorithm,and chaos optimization(CO)is added in the optimization-seeking process to prevent the particles from falling into local optimal solutions,and the improved particle swarm algorithm is used for Elman model parameter-seeking to solve the problem of difficult selection of Elman model parameters.The average color difference of the ACOPSO-Elman model characterization is 1.9247 ΔE∗ab and the maximum color difference is 5.1252 ΔE∗ab,which achieves better results in the characterization accuracy,as verified by simulation experiments and compared with BP and Elman neural network models.
Neural networkLCDColor characterizationPSO algorithmAdaptive adjustment function
NETL
NSTL
万方数据
