【作者】 陈娟； 张勇；

【Author】 Chen Juan Zhang Yong (School of Control Science and Engineering University of Jinan Jinan 250022)

【机构】 济南大学控制科学与工程学院；

【摘要】 由于BP网络在模拟非线性曲线方面的优势,其被用于清晰化研究中.但由于权值的影响,在训练过程中会出现各种问题.为解决这一问题,本文在原有的利用BP网络进行清晰化研究的基础上,提出了一种基于遗传算法的清晰化方法.该方法在BP网络训练之前,首先用遗传算法估计出权值的初始范围,从而避免了网络训练过程中权值对训练时间、训练速度的影响.通过改进后网络的训练结果与传统BP算法结果的对比,可发现,改进后的方法可以有效减少训练时间,并改善视频质量.更多还原

【Abstract】 As BP network’s priority in the simulation of non-linear curve areas,it is used in clearing of fuzzy videos.However,due to the impact of weight in the training process,various problems arise.To solve this problem,this article presents a clearing method based on genetic algorithm,which makes use of the original BP network’s clearing research.Prior to BP network training,the method firstly using genetic algorithms to estimate the range of the initial value,thus avoiding the impact of weight training time,the speed of training during the process of network training,.By comparing the training results of the traditional BP algorithm and the paper’s results,we can find that the improved method can effectively reduce training time and improve the video quality.更多还原