摘要
针对车牌字符在车牌图象退化时识别率较低的问题,提出一种基于神经网络集成的车牌字符识别方法。基于小生境遗传算法在提高进化的局部搜索方面的良好性能来动态构建个体网络差异性大的神经网络集成,进而提高整个集成系统的泛化能力。将该方法应用于车牌字符的识别,实验结果表明,该方法能有效地生成差异度较大的个体网络,得到的神经网络集成能有效提高车牌字符的识别率。
Aiming at the current problem that license plate character recognition is at low recognition rate when the license plate image quality degrades, this paper presents a license plate character recognition method which is based on neural network ensemble. Using niche technique's good performance in improving local search of evolution to dynamically construct individual networks with diversity, and to improve the generalization ability of the neural network ensemble. At last, the method is applied in the license plate character rec- ognition. The results show that the method can generate individual networks with greater different degrees and the neural network ensemble it generates can effectively improve the rate of license plate character recognition.
出处
《安徽广播电视大学学报》
2012年第2期116-120,共5页
Journal of Anhui Radio & TV University
基金
安徽省高校优秀青年人才基金项目(2012SQRL230)
关键词
神经网络集成
小生境
遗传算法
车牌识别
neural network ensemble
niche
genetic algorithm
license plate recognition