视频中目标运动参数的快速提取技术研究  

Research on Fast Extraction of Motion Parameters in Video

作　　者：乔宏章[1] 刘志荣 张军[3] 

机构地区：[1]中国电子科技集团公司第五十四研究所,河北石家庄050081 [2]中国电子进出口总公司,北京100036 [3]天津工程师范大学,天津300222

出　　处：《无线电工程》2014年第9期16-20,共5页Radio Engineering

基　　金：国家部委基金资助项目

摘　　要：各类飞行器、机动车的模拟驾驶环境的建立,需要从视频影像中进行运动目标检测并且提取出图像运动坐标系下的对应运动参数。由于在凝视拍摄过程中,摄像机以运动目标为视场中心追随运动目标,此时的运动影像可以看作是具有全局运动特征的图像。从驾驶环境模拟的角度提取视频影像中主体目标的运动特征和参数,提出了基于小波分解的块平均匹配方法,提高了参数提取的准确性和处理速度。仿真实验验证了算法的有效性。It is necessary to detect objects and extract the motion parameters in video for the establishment of virtual driving environment of various aircrafts and vehicles. Since the camera follows the moving object as the view field center during the imaging process of moving objects in battlefield,the battlefield video image can be considered as an image with a global motion feature. Motion characteristics and parameters are extracted from the viewpoint of battlefield environment simulation,and a method based on wavelet hierarchical block matching is presented. The accuracy of parameter extraction and the processing speed are improved. The simulation results show that the algorithm is efficient.

关 键 词：虚拟驾驶 运动参数估计 块匹配 全局运动估计 小波分解 

分 类 号：V556[航空宇航科学与技术—人机与环境工程]