复杂背景下基于车牌特征的车牌粗定位实现

本文基于车牌灰度特征和几何特征实现了车牌识别系统(LPRS)中车牌的粗定位。采用直方图Ostu方法选取阈值进行二值化、采用3*3大小的十字形窗口进行中值滤波,分别用水平投影和垂直投影法寻找车牌的左右边界和上下边界,结合图像灰度序列与单位矩形序列的互相关函数定出车牌的位置,为以后车牌进一步精确定位以及字符识别打下基础。

Realization of the license plate location based on the characters in Complex background

This article realized the license localization in the car license recognition system (LPRS)based on the car license gradationcharacteristic and the geometry characteristic. Using the histogram Ostu method to select a threshold in binarization, using the 3*3size- cross- window value filter, using horizontal project and vertical project to seek the car license boundary, then through the mutualcorrelation function of union picture gradation sequence and the unit rectangle sequence decide on the car license the position, build-ing the foundation for later car license further precise localization and the character recognition.