基于BP神经网络的货车车号识别方法研究  

Research on method of vehicle number recognition based on BP Neural Network

作　　者：杨绍华[1] 李建勇[1] 王恒[1] 

机构地区：[1]北京交通大学机械与电子控制工程学院,北京100044

出　　处：《铁路计算机应用》2007年第12期4-7,共4页Railway Computer Application

摘　　要：针对货车车号的字体不规则、单字有断裂的特点,提出基于图像预处理的动态字符分割和提取算法,准确地对车号图像进行车号区域提取和车号单字分割。利用Hilditch细化算法对单字进行细化,提取出能表征数字的结构特征向量。设计出BP人工神经网络,数字的结构特征向量作为BP网络的输入,用经过训练的BP网络进行货车车号的识别。The vehicle number had the interrupted character and a single character may have ruptures. According to these characters, a dynamic character partition and acquisition algorithm was presented, so it could be accomplished vehicle number area location and single character partition. Based on these, Hilditch thinning algorithm was used to thin the character and pick up the structural character vectors. At last, BP neural network was designed and the vectors were used as network＇s inputs. Through the trained BPNN, the vehicle number recognition was implemented.

关 键 词：字符分割 车号识别 BP神经网络 学习算法 

分 类 号：U269[机械工程—车辆工程] TP183[交通运输工程—载运工具运用工程]