一种基于HSV颜色空间与小波能量特征融合的车牌定位算法

Vehicle License Plate Location Algorithm Based on Feature Fusion of HSV Color Spaces and Wavelet Energy

文中针对彩色汽车图像牌照定位率不高和速度慢等问题,提出了基于HSV颜色空间与小渡能量分析相融合的车牌定位方法.利用固定底色牌照的色相、亮度和饱和度值分布区间不同的特点,完成对大多数黄蓝底色牌照的快速定位.对于车牌底色与车身颜色相近无法一次完成定位的操作,进行图像二次定位处理,利用车牌区域小波高频能量分布特点从背景图像中提取车牌区域.实验结果表明该方法比单一使用HSV空间进行定位的算法准确率高,比其它小波定位算法实现速度快,并且具有较好的鲁棒性.

This paper proposes a feature fusion method based on HSV color spaces and wavelet energy analysis to solve the problems of existing vehicle license plate locating algorithm such as low locating rate of colored car image license and slow speed. Firstly, it uses the characteristics of hue, brightness and saturation distribution range of fixed background license plate to improve locating speed. Secondly, it uses the dis- tribution features of wavelet high frequency energy to locate color vehicle image that license plate back- ground is similar to vehicle body color. The experimental results demonstrate that the method is higher accuracy than using single HSV algorithm, locating speed is quicker than other wavelet algorithms, and has better robustness.