Abstract: In order to overcome the problem of slow execution in using SURF algorithm to locate license plate accurately, a novel car license plate location algorithm based on multiple features and SURF algorithm is proposed. The algorithm roughs the candidate region through HSL color space and uses geometric characteristics and textural feature to filter the candidate region and marks the candidate region by the set rules. Then it extracts the label candidate region in order and uses SURF algorithm to locate the plate precisely in this region. The experimental results show that the SURF algorithm has better effect compared with SIFT algorithm in the execute time, and it can meet the needs of the real time.

Key words: license plate location, HSL color space, geometric characteristics, textural feature, Speeded Up Robust Feature（SURF） algorithm

摘要： 针对利用SIFT算法进行车牌精确定位时执行时间较长的问题，提出一种基于多重特征和SURF算法的车牌定位算法。在HSL颜色空间得到车牌候选区域，结合车牌几何特征与纹理特征筛选候选区域并按设定的规则标号;按序提取车牌标号候选区域，并用SURF算法对候选区域精确定位。实验结果表明，SURF算法与SIFT算法相比，在定位准确率相同情况下减少了运行时间，能满足实时性需要。

关键词: 车牌定位, HSL颜色空间, 几何特征, 纹理特征, 快速鲁棒特征（SURF）算法