桥区运动船舶的检测和跟踪  

Detection and Tracking of Moving Ship in Bridge Area

作　　者：张文涛[1] 谭毅华[1] 余晋刚[1] 田金文[1] 

机构地区：[1]华中科技大学多谱信息处理技术国家级重点实验室,武汉430074

出　　处：《计算机与数字工程》2011年第10期67-71,156,共6页Computer & Digital Engineering

基　　金：国家自然科学基金(编号:50909079);教育部专项基金(编号:109109)资助

摘　　要：基于桥区的场景特点,提出一种基于自适应纹理特征的运动船舶目标检测跟踪算法。算法采用基于改进的局部二值模式(LBP)纹理特征取代亮度特征,在此基础上建立可实时更新的LBP背景,以背景模型方法提取船舶运动目标前景,并以简单的近邻方法实现对其跟踪。实验表明该方法通过目标检测与跟踪可实现对船舶运动的监控,算法准确性好,精度高,可为桥梁安全保护提供有效的技术支撑。A moving ship detection and tracking method based on the adaptive texture features of the bridge area is presented.The background model is established by replacing brightness of LBP texture features to establish the background model,and updated for every coming frame in real time.Foreground areas are extracted as moving ships by adaptive threshold.Finally,the moving ships are tracked using simple nearest neighboring method.Experiments show that it＇s accurate and precise to detect and track moving ship in the actual video.Therefore,as a crucial technique,it is able to be applied to the bridge protection system.

关 键 词：运动目标检测 运动目标跟踪 局部二值模式 船舶监控 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]