摘要
传统频域内的运动目标检测算法通常需要设计三维时空滤波器组来实现,存在待选参数多、滤波器设计难度大等问题。文中提出了一种二维频域内的运动目标检测方法,通过对动态图像的行列分解,将三维频域内的运动检测问题转化到两组二维频域内进行,从而降低了滤波器设计的难度。给出了一种提取主运动能量的自适应滤波算法,通过剔除背景和噪声的频率成分,有效地检测出运动目标。仿真结果表明,该方法可有效降低背景配准误差和噪声的影响。
Usually, 3D spatio-temporal filter bank is designed in 3D frequency domain to detect moving target, but the design method of the filter is difficult because of too more parameters and design difficulty. A novel moving target detection algorithm in 2D frequency domain is proposed. By decomposing the dynamic images into row and column sequences, the 3D frequency domain is translated into two groups of 2D frequency domain, and the filters design becomes easier. Then an adaptive filter is designed to extract the motion energy and pick out the frequency components of static background and noises, and the moving targets can be detected effectively. Simulation results show the robustness against image noises and matching errors.
出处
《红外与激光工程》
EI
CSCD
北大核心
2007年第1期135-138,共4页
Infrared and Laser Engineering
基金
国家自然科学基金资助项目(60674043)
关键词
动态图像分析
运动目标检测
成像制导
成像跟踪
Dynamic image analysis
Moving target detection
Imaging guidance
Imaging tracking
