Abstract: In view of the numerous methods to recognize license plate in image, most use optical character recognition based method. And confounding factors such complex scene, occlusion and background clutter, make license plate recognition in the real scene be far away from matured. To this end, a SIFT matching based method for license plate recognition is proposed in this paper. The interpolation based super-resolution method is used as preprocessing for image reconstruction. And a license plate segmentation method is proposed by using shape feature. Finally, SIFT matching method is used for license plate recognition. Experimental results show that the proposed method locates and recognizes license plate in image accurately.

Key words: Scale Invariant Feature Transform（SIFT） feature matching, license plate recognition, intelligent transportation system

摘要： 针对智能交通系统中基于视觉的车牌识别中存在的依赖于光学字符识别以及在复杂环境下准确率低的问题，提出了基于SIFT特征匹配的车牌识别方法。通过基于插值的超分辨率图像重建方法对车牌图像进行预处理，基于轮廓特征对车牌进行定位，通过SIFT特征匹配的方式，利用模板库中的车牌字符模板对车牌进行定位验证以及字符识别。实验结果表明该方法能有效提高车牌识别的效率。

关键词: 尺度旋转不变特征描述算子（SIFT）特征匹配, 车牌识别, 智能交通系统