【作者】 方江波；

【导师】 徐建闽；

【作者基本信息】 华南理工大学 ， 控制理论与控制工程， 2010， 硕士

【摘要】 “车牌识别系统”是智能交通系统的一个重要组成部分,对提高交通系统的车辆监控和管理的自动化程度具有很高的实用价值和经济价值。本文围绕基于TMS320DM642的车牌识别系统这一主题展开研究。首先简述了车牌识别系统的原理,通过与PC机车牌识别系统的对比,提出了基于TMS320DM642车牌识别系统在体积、重量、能耗、速度等方面的明显优势,研究如何在DSP硬件上实现车牌识别系统。接着介绍了TMS320DM642的硬件开发平台,包括TMS320DM642芯片的结构、EVM、指令系统和软件开发环境。然后,详细论述了完整的车牌识别系统中每个部分的算法。采用背景差分的方式对运动车辆进行视频检测,获取单帧车辆图片;提出了一种改进Prewitt算子的边缘检测方法,将它与统计投影相结合来实现车牌定位;分割垂直投影经过小波阈值去噪处理的车牌图像,将分割得到的车牌字符作归一化处理;利用BP神经网络进行字符识别,获取车牌信息。最后,通过线程通信的方式完成车牌识别系统的DSP软件模块化开发,将识别结果在Dot Net平台开发的PC端软件上显示与存储。并对该系统进行代码优化、存储器优化等,使得DSP车牌识别软件运行效率更高、识别速度更快。更多还原

【Abstract】 The LPR (License Plate Recognition) system is a significant part in ITS (Intelligent Transportation System), which increases Transportation Vehicle Monitoring System and Management Automation, and shows high practical value and economic value.This paper studied the LPR system based on TMS320DM642 chip. First of all, outlines the principles of LPR system, which based on TMS320DM642 has advantages in size, weight, power consumption and processing speed comparing with PC LPR system. Give us a solution of a LPR system based on DSP chip.Then to introduce the TMS320DM642 hardware development platform, including the TMS320DM642 chip structure, EVM, command systems and software development environment.Next to discuss each part of the algorithm of the LPR system in detail. Using the background difference to make the detective video of moving vehicles for single frame image of the vehicle, an improved Prewitt edge detection operator method, combining it with statistical projection to achieve plate positioning, vertical projection of Wavelet threshold denoising license plate image for segmentation, to get the license plate character segmentation for normalization, and using BP neural network for character recognition, to obtain license plate information.Finally, modular DSP software development of LPR system is accomplished through thread communication. The recognition results in the development of the Dot Net platform, client software on the PC display and storage. And the system code optimization, memory optimization, makes the LPR system based on DSP chip to run more efficient and faster recognition speed.更多还原