摘要
提出利用基于图像边缘特征的相位相关简化方法来进行空间目标平移运动的检测。为了能有效抑制噪声和亮度变化的影响,在传统互相关方法的基础上提出一种先提取目标的边缘特征,再利用傅里叶平移特性进行归一化相位相关处理的检测方法,从而实现两幅相似图像间的相对平移矢量的准确估计,它较比时域方法和基于灰度信息的互相关方法能获得更准确的检测结果和更强的噪声抑制能力。同时,为满足系统实时性的要求,还采用了将二维图像映射为一维信号的简化方法,将程序执行时间缩短了4倍。仿真实验结果显示,利用边缘特征的简化方法对于受到严重噪声干扰的空间运动目标具有很好的检测结果,进一步证明了算法的有效性和实用性。
In this paper, it is presented that a reduced method of phase correlation based on edge character is used to identify the shift vector of spacial moving object in the frequency domain. To effectively restrain noise and influence of image illumination, a normalized phase correlation method based on well-known Fourier shift property is applied after extracting edge character, it can accurately estimate the motion of abject by identifying the displacement between two similar images. It can obtain more exact matching result and better ability of restraining noise than traditional cross correlation method and arithmetic in the spatial domain. To satisfy the requirement of real lime system, a reduccd method is presented, which make image of two dimensions become signal of one dimension by projection, by which prugram nmtime is cut short fourfold. Simulation experiment prove that this method give a better detect result for the spatial object seriously polluted by noise, which demonstrates the utility and practicability of this method.
出处
《宇航学报》
EI
CAS
CSCD
北大核心
2006年第5期1107-1112,共6页
Journal of Astronautics
关键词
运动估值
频域
傅里叶变换
相位相关
边缘提取
Motion estimation
Frequency domain
Fourier transform
Phase correlation
Edge detecting
