论文部分内容阅读
提出了一种从频率域出发,估计出运动物体在空间域上位移的算法.利用傅里叶变换的自配准性质和位移特性,用极坐标形式下连续图像相位谱的差直接估计运动目标的位移,并进一步推广用于解决高速公路上视频交通监测中车速的非接触式、动态和高精度测量.与传统的寻找迪拉克峰值的方法相比,该算法先求出相位谱的差,再用其周期数来估计位移的算法简单明了,由于其无需重新变换回空间域,从而节省了处理时间,具有更好的实时性,且精度高,实验证明其分辩能力不低于一个像素,其平均测速精度不低于95%.
An algorithm is proposed to estimate the displacement of moving objects in the space domain from the frequency domain. By using the self-registration property and displacement characteristics of Fourier transform, the moving target is directly estimated by the difference of the continuous image phase spectrum in polar form And further popularize the non-contact, dynamic and high-precision measurement to solve the speed of video traffic monitoring on the highway.Compared with the traditional method of finding Dirac peak, this algorithm first obtains the difference of phase spectrum, And then use its number of cycles to estimate the displacement of the algorithm is simple and clear, because it does not need to re-transform back to the spatial domain, thus saving the processing time, with better real-time, and high precision experiments prove that their resolution is not less than one pixel, The average speed accuracy of not less than 95%.