Transverse velocity estimation based on Wigner-Hough transform

Bai Ye-Chao,Zhang Xing-Gan,Tang Lan

2010-01-01

Abstract:Velocity is an important character of moving objects,which is widely used in target tracking,target recognition and so on. The velocity can be decomposed into radial part and transverse part. The radial velocity of an object can be easily estimated by means of the Doppler shift,which manifests itself as the one-order phase of the echo. However,the one-order phase of the echo does not contain any information about the transverse velocity. The transverse velocity can be estimated by the high-order Doppler effect of the moving object. The time-frequency characteristic of the echo of transverse moving targets is analyzed in this paper,and a conclusion that the Doppler frequency varies linearly with respect to time is drawn. Also,the analytical expression for the transverse velocity is deduced. An algorithm based on Wigner-Hough transform is proposed to estimate the transverse velocity. The proposed algorithm transforms the echo signal into time-frequency plane by using Wigner-Vile distribution,then estimates the slope of Doppler frequency-time line with the help of Hough transform,and finally calculates the transverse velocity. The estimation accuracy is also analyzed. Simulations are carried out to demonstrate the effectiveness of the proposed algorithm. Simulation results verify that the echo frequency is a linear function of time. Track Fitting algorithm and Phase Difference algorithm are simulated together to be compared with the proposed algorithm. The proposed algorithm performs well in the presence of noise and basic frequency offset,and has a better accuracy than other algorithms. Track Fitting algorithm is not suitable for signals with nonzero basic frequency. The performance of Phase Difference algorithm severely decreases when signal-to-noise ratio is low. Constant radial acceleration also leads to a linearly varying frequency,which affects the transverse velocity estimation. However,there are very few targets which do uniformly accelerative movement for a long time. The jitter of radial acceleration interferes with the transverse velocity estimation. The simulations with radial acceleration interference show that the proposed algorithm performs much better than other algorithms,and has a better robustness.