节点文献
自动车牌定位及字符分割研究
【作者】 潘仁龙；
【导师】 王林；
【作者基本信息】 贵州民族学院 ， 概率论与数理统计， 2010， 硕士
【摘要】 智能交通是当前交通管理发展的主要方向,汽车牌照识别技术则是智能交通系统的核心,在城市道路管理、港口、机场、高速公路和停车场等项目管理中占有重要地位。车牌自动识别系统分为车牌定位、车牌字符分割和字符识别三个部分,其中车牌定位和字符分割是核心技术,直接影响系统的识别率和速度。本文针对车牌定位和车牌字符分割进行了较系统的研究。首先,以图像处理技术为基础,对采集到的车牌图像进行预处理,得到质量较好的汽车图像,这为后续的车牌定位及字符分割处理打下良好的基础;其次,根据车牌图像具有丰富的纹理和车牌区域为矩形且边缘互相平行等特征,提出了一种基于方向场和车牌边缘信息的车牌定位方法,并针对车牌图像存在的倾斜和变形,提出了一种基于方向场的车牌水平倾斜校正和基于单个字符投影的竖直变形校正方法;最后,利用统计的方法,对字符特征进行分析,采用垂直投影法对字符进行分割。在上述算法的基础上,我们用Visual C++完成一个实验系统,通过对200多幅车辆图像进行测试。实验结果表明,对各种不同大小、光照条件不同的图像,建议方法都能准确地定位车牌和分割字符。这些为下一步的字符识别及完成车牌自动识别系统的工作奠定了基础,具有较好的理论研究意义和实际应用价值。
【Abstract】 The intelligent transportation is the main direction of the development of transportation management. License plate recognition (LPR) technology is a key of intelligent transportation systems. It is important in the project management of urban road management, ports, airports, expressway and parking area, etc. Usually, the LPR system consists of three main parts: license plate localization, character segmentation, and character recognition. The license plate localization and character segmentation are the most crucial stage. It result has an important impact on the accuracy and speed for the LPR system. The propose algorithm with research to the license plate location and character segmentation.Firstly, the vehicle license plate image is preprocessed with image processing technology to get better quality vehicle image. It has built a good foundation for license plate localization and character segmentation. Secondly, the license plate upper and lower borders are parallel each other. There are rich edge and texture information. We present an algorithm based on orientation field and edge information for location of license plates. Furthermore, we present a new algorithm based on orientation field for horizontal skew correction and an algorithm based on single-character projection for vertical distortion correction. Finally, character feature is analyzed with statistic method, and license plate character is segmented by vertical projection method.Based on the above algorithm, we develop practical experimental system with Visual C++ and it is tested with 200 various vehicle images. Experimental results show that proposed method can accurately localization and character segmentation various license plate of different size and illumination. The research is an important foundation for next character recognition and LPR system completing, and it has both better the theoretical research significance and applied value.