Road safety: Research and reality 6th Reuben Smeed memorial lecture

Research has had a major influence in advancing road safety in Britain over more than 60 years. Lessons learnt in the early years provide a model for effective conduct of research of high quality to aid and influence practice and policy making. As the control of road safety research has become more widespread over the last decade and effort more diffused, problems have arisen which militate against the appropriateness and successful application of findings. Conflicts between researchers, practitioners and policy makers stem from lack of mutual understanding of their roles, the need to balance scientific rigour against the urgency of government and practice, and inadequate coherent communication.The potential for road casualty reduction in Britain has been addressed at intervals since the early 1960s. Assessment of the latest trends in casualties, set against current activities to achieve target reductions, shows the need for a further update to redefine priorities. For future advancement of road safety, a 20-point plan of activities is suggested. Technical, social, economic and institutional difficulties, of varying degrees, will need to be overcome to achieve the objectives of the plan. Most importantly, the academic and professional standing of road safety workers needs to be raised.     

基于图像处理的内河船舶目标提取与特征值计算

内河航行船舶进行目标提取及特征值计算,是船舶航行监控及船舶交通流参数测量的基础。文中主要介绍:1将背景差分法与大津法相结合,提出了1种新的船舶目标提取方法,对不同天气、不同角度下内河在航船舶目标进行提取;2提出了基于圆环划分的特征值算法,通过此算法计算所得目标船舶的特征值。验证结果表明,基于背景差分结合大津法可实现船舶目标提取,所得船舶轮廓较为清晰,并具有一定天气适应性;在特征值计算实验中,正拍图像的特征值1均大于0.95,而侧拍图像特征值1均小于0.5,可区别不同拍摄角度;各种船舶目标在进行旋转、平移及缩放后,特征值基本保持不变。因此,文中算法所得到的特征值具有旋转、平移及缩放的不变性,并可以适应在不同角度下摄像机所拍摄的要求。     

复杂背景下海天线的连通域检测法

海天线的检测对于划分海空背景、识别红外舰船目标具有重要作用.本文提出一种基于线段表连通域原理的海天线检测方法.首先去除原始红外图像的强水纹干扰和背景干扰,再通过梯度运算和最大值滤波找出海天线的位置.最后基于线段表连通域原理对二值化后的图像进行检测,以提取海天线的非线性边缘信息.实验表明,该方法的检测精度高,而且当海天线上存在舰船时,不会对舰船轮廓产生影响,克服了Hough变换法只能检测出直线的缺点,具有较强的实用性.