车牌自动识别中的字符特征提取  

Character feature extraction for car plate recognition

作　　者：董玲娇[1] 

机构地区：[1]温州职业技术学院电气电子工程系,浙江温州325035

出　　处：《机电工程》2008年第9期106-108,共3页Journal of Mechanical & Electrical Engineering

摘　　要：为提高车牌字符识别率,根据车牌字符特性提取了笔画的4个复杂性指数,以及面积、复杂度、占空比共7个典型的形态学特征,并对其进行了归一化处理。研究结果表明,该方法的结构特征明显、运算量小、速度快,识别准确率和效率相对于其他特征提取方法有很大提高。Aiming at improving car plate character recognition rate, seven typical morphology characteristics were extracted and normalized according to the distributed characteristic of characters. They were four complex exponents, area, complexity, and duty ratio. Experimental results show that the method has the advantage of distinguishing architectural feature and less computational, it also improves the recognition accuracy, speed and efficiency compared with other feature extraction methods.

关 键 词：特征提取 复杂性指数 归一化 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]