基于彩色形态学与支持向量机的实时汽车牌照识别

Real-time Vehicle Plate Recognition Supporting Vector Machine and Based on Color-scale Morphology and

车牌识别系统是智能交通系统的重点研究方向。本文在对现有技术的研究和分析基础上,提出了一个实时车牌识别系统,该系统由五部分组成:利用彩色形态分割与判断边缘颜色对进行彩色边缘检测,生成连通域进行车牌定位,倾斜矫正,字符分割,以及利用多类支持向量机进行字符识别。通过对大量的实践验证,该系统克服客观条件和参数的影响,具有良好的实时性和更大的应用范围。

Vehicle license plate recognition system is animportant application field in intelligent transportation system. A realtime vehicle plate recognition system can be divided into five sections: color edge detectionby using color-scale morphological segmentation and color-pair discrimination,connect region location, tilt correction, character segmentation and recognition using Multi-class Support Vector Machine. Experimental resultsshowed that without the influence of external conditions and parameters , the proposed system had good real-time performance and large applicable range.