论文部分内容阅读
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。
The moving object detection algorithm in the traditional frequency domain usually needs to design a three-dimensional space-time filter bank to realize, there are many problems to be selected, and the difficulty of the filter design is large. In this paper, a moving target detection method in two-dimensional frequency domain is proposed. By decomposing the moving image matrix, the motion detection problem in the three-dimensional frequency domain is transformed into two groups of two-dimensional frequency domain, which reduces the difficulty of the filter design. An adaptive filtering algorithm for extracting the main motion energy is given. The moving target can be effectively detected by eliminating the frequency components of background and noise. Simulation results show that this method can effectively reduce the impact of background registration error and noise.