Gait Variations in Human Micro-Doppler

INTL JOURNAL OF ELECTRONICS AND TELECOMMUNICATIONS, 2011, VOL. 57, NO. 1, PP. 23–28 Manuscript received January 19, 2011; revised February 2011. DOI: 10.2478/v10177-011-0003-1 Gait Variations in Human Micro-Doppler Dave Tahmoush and Jerry Silvious Abstract—Measurement of human gait variation is important for security applications such as the indication of unexpected loading due to concealed weapons. To observe humans safely, unobtrusively, and without privacy issues, radar provides one method to detect abnormal activity without using images. In this paper we focus on modeling the characteristics of human walking parameters in order to determine signature differences that are distinguishable and to determine the variability of normal walking to be compared to armed or loaded walking. We extract micro-Doppler from motion-captured human gait models and verify the models with radar measurements. We then vary the model to determine the extent of normal micro-Doppler variation in multiple dimensions of human gait. We also characterize the ability of radar to determine gender and suggest that alternative views to the frontal view may be more discriminative. Keywords—Radar, human, gait. I. I NTRODUCTION F OR observing humans, radar has advantages over other sensors. Radar signals can penetrate clothing, preventing disguise from being effective, while not compromising individual privacy. Using radar to determine unexpected loading, and thus to identify individuals trying to smuggle weapons or other items through a security checkpoint, is of interest for security applications. Understanding the variability of normal human motion as viewed by the radar can determine the capabilities and limitations of this type of device in determining loading accurately. Detailed radar processing can reveal characteristics of the walking human. The different parts of the human body do not move with constant radial velocity; the small microDoppler signatures are time-varying and therefore analysis techniques can be used to obtain more characteristics [1], [2]. The modulations of the radar return from arms, legs, and even body sway are being studied [3]–[5]. We analyze these techniques and focus on modeling human body motion to simulate the variations. The Doppler information measured by a radar arises from target motions. If we denote the target position by P (T ), where the coordinates x and y are functions of slowly varying time T and the origin is the radar:   x(T ) P (T ) = (1) y(T ) then the instantaneous radial target speed is given by vr (T ) = ~r(T ) d P (T )  dT |~r(T )| (2) This work was supported in part by the U.S. Department of Defense. D. Tahmoush and J. Silvious are with the US Army Research Laboratory, 2800 Powder Mill Road, Adelphi, MD 20783, USA (e-mails: {david.tahmoush, jerry.silvious}@us.army.mil). where ~r(T ) stands for the vector between the radar and the target. The resulting Doppler frequency shift Fd is then: 2vr (T ) 2Ft Vr (T ) = (3) λ c where Ft is the frequency of the transmitted signal, λ is the wavelength, and c is the speed of light. The equation for computing the non-relativistic Doppler frequency shift of a simple point scatterer moving with speed v with respect to a stationary transmitter is: Fd (T ) = 2v cos θ cos φ (4) c where θ is the angle between the subject motion and the beam of the radar in the ground plane, and φ is the elevation angle between the subject and the radar beam. This assumes that the radar itself is stationary. For complex objects, such as walking humans, the velocity of each body part varies over time as the person walks. The radar cross-section of various body parts is also a function of aspect angle and frequency. The Doppler of a moving vehicle is similar to a point scatterer, but humans have a larger spread of velocities due to their bipedal motion. A short-time FT (STFT) is one way to explore the slowtime dependent behaviour of the Doppler spectrum by doing a Fourier transform over a small window in time, then sliding the window [6]. This avoids the loss of time information that occurs when applying a Fourier transform. The continuous form of the STFT is: Z ∞ ST F T (x(t)) = X(τ, ω) = x(t)w(t − τ )e−jwt dt (5) Fd = Ft −∞ where w(t) is the window function. Because human microDoppler varies slowly with time, we employ STFTs of the IQ radar data. The length of time used in the STFT is called the dwell time or coherent processing window, and this determines the resolution in Doppler frequency that can be measured. This can partially be overcome by super-resolving methods. The spectrogram is the square modulus of the STFT and is then: Spectrogram(τ, ω) = 10 log10 |X(τ, ω)|2 (6) Which is often used to display micro-Doppler data in decibels, as is done for the images in this paper. Much of the analysis in this paper makes use of spectrograms for the display of micro-Doppler phenomenology. We perform simulations of the human gait and verify them with radar measurements. We break down the radar spectrogram into its components based upon simulated and measured human signatures. We model the variation to be expected when measuring human micro-Doppler signatures and compare them to the measured variations. We then analyze the capability of detecting gait variation due to loading as a security technology. 24 D. TAHMOUSH, J. SILVIOUS II. S IMULTATION M ETHOD Simulation of the human gait has been performed by many researchers, often with the goal of improving animated movies [7]–[13]. Here we are taking the extensive research on human gait and animation and using it to model the expected Doppler shifts measured over time by a radar system. We started with the measurements made in [14]. Twenty men and twenty women whose ages ranged from 20 to 38 years with an average age of 26 years had their motions captured on video and extracted, then their characteristics analyzed. The resulting motion information was extracted, and then animated. We took the animated gait and extracted the micro-Doppler velocities that would be created by differentiating the motions using a point-scatterer model for each separate part. We neglected obscuration for these simulations because they were limited to frontal-view, and we used a metallic skin approximation to simplify the calculations. The simulated micro-Doppler motions for different body parts are shown in Figure 1. These are calculated from the actual motions of the model and are calculated at 17GHz. The resulting spectrogram and a comparison with measured radar data is shown in Figure 2. The scaling for the images was set at a 2m/s foot swing max in order to simplify the comparison of images to demonstrate the variability of the human gait as viewed by the radar. The stride rate is also held fixed to simplify comparisons. We also do not simulate noise in the models. Highly accurate meshmodeled simulations of the human micro-Doppler signature have been done [15] but not with studies of the var (...truncated)