道路车辆的信息检测及视觉理解

The vehicle's intelligent visual ssurveillances is the core issue of intelligent transportation systems from manual analysis to the development of the intelligent analysis. And vehicle information detection, visual analysis and understanding are the bottleneck of vehicle intelligent visual surveillance. The study not only has broad application, but also has important theoretical research value. This project intended to the forefront of vehicle information detection and visual analysis. Such as vehicle license plate and vehicle detection, vehicle attitude evaluation function and optimization problems, close to the real motion of the vehicle motion model, the vehicle posture prediction, the occluded object detection, the determination of the visible area, Multi-vehicle mutual occlusion, etc. Propose some new algorithms and new ideas, Completion experimental system containing these cutting-edge issues, preparing an experimental system, which including cutting-edge issues, Enrich and improve the vehicle visual monitoring theory, High-quality papers will be published in journals (index for SCI, EI) and international conferences. Some results will apply for a patent. Through this project, evaluation and optimization of vehicle attitude, the vehicle posture prediction, occlusion handling and other research has reached the international advanced level.

路面车辆的智能视觉监控是智能交通系统从手工分析向全面智能化分析发展的核心问题；而车辆的信息检测及视觉分析和理解是路面车辆智能视觉监控中的瓶颈问题。这些研究不仅具有广泛的应用前景，而且具有重要的理论研究价值。本项目拟对车辆信息检测及视觉分析的一些前沿问题，如车辆的车牌车型检测、车辆姿态评价函数的建立和相关优化问题的求解、与车辆的真实运动接近的运动模型、车辆姿态的预测、遮挡出现的检测、可见区域的确定、多车辆相互遮挡等，开展研究，提出一些有效的新算法和新思想，完成一个包含这些前沿问题的实验系统，丰富和完善路面车辆的视觉监控理论，在国内外SCI、EI收录刊物及重要的国际学术会议上发表一批高质量的学术论文，部分成果有望形成专利。通过本项目的研究，在车辆姿态评价与优化、车辆姿态的预测、遮挡处理等研究上达到国际先进水平。

随着车辆保有量不断攀升，道路监控的难度和重要性显得越来越突出。而车辆的信息检测及视觉分析是路面车辆智能视觉监控中的瓶颈问题。本项目围绕道路车辆信息检测及视觉分析的一些前沿问题（如车辆的车牌、车型、车标等信息），采用数字图像处理、模式识别和计算机视觉的方法，通过对监控摄像机拍录的图像序列进行自动分析，对固定场景中活动目标进行定位、识别和跟踪，获取运动车辆的信息。通过项目的研究，提出了一些有效的算法，完成了1个实验系统，在国内外学术刊物及国际学术会议上发表了29篇学术论文和获得2项软件著作权。