【作者】 范钢锋；

【导师】 李金伴；

【作者基本信息】 江苏大学 ， 系统工程， 2002， 硕士

【摘要】 本论文对复杂背景情况下机动车辆牌照的识别作了深入研究。在深入掌握和理解图像处理和软计算等相关知识的基础上,针对识别过程的各个阶段,包括牌照分割、字符切分、LVQ2神经网络的字符识别和LVQ神经网络权值的优化等,提出了基于连通成份标记和复合极值形态学滤波的牌照自动分割算法,探讨了用LVQ神经网络对牌照上的字符进行识别的方法,并将基因算法引进了LVO神经网络的权值优化,取得了良好的效果。本文的方法对其它图象处理问题也有较高参考价值。论文主要内容包括:1)图像处理和识别基础2)牌照分割3)字符识别前预处理4)LVQ2神经网络模式识别5)遗传算法在LVQ中的运用更多还原

【Abstract】 The thesis studies on automatic recognition of vehicle plate from complicate background. Based on diving into image manipulation, soft algorithm and other theories related, for each stage of the object recognition, including extraction of license plate, segmentation of character, character recognition by LVQ2 NN and the optimization of LVQ NN weight, the thesis puts forward the method of automatic extraction of license plate based on sequential mark of regions and compound extremum morphology filter, researches into text recognition by NN and optimizes the LVQ weight by Genetic Algorithm.. These two methods are effective. The methods of this thesis are also useful to resolving other problems of image processing.The highlights of the thesis include:1) image manipulation and theory basis of recognition2) automatic extraction of license plates3) pretreatment before recognition4) character recognition by LVQ25) Genetic Algorithm applied to LVQ更多还原