Microwave Imaging of Non-Rigid Moving Target Using 2D Sparse MIMO Array

Zhanyu Zhu,Lei Kuang,Feng Xu
DOI: https://doi.org/10.1109/access.2019.2945968
2018-01-01
Abstract:This paper develops a microwave/mmw (millimeter-wave) imaging method for moving objects of non-rigid body which can be used in the heavy traffic scenarios. This method is proposed with a 2D sparse MIMO (multiple-input multiple-output) array composed of two orthogonal linear arrays. In comparison to the RX antennas, a smaller number of TX antennas is used to decrease the data acquisition time of this system when using TDM (time division multiplexing) signal transmitting technique. The 2D sparse MIMO array is deployed with wide-band signals transmitted to reconstruct the backscattering coefficient in 3D space using synthetic aperture imaging technique. Then a series of synthesized 3D snapshot images are generated during the motion of the target. The non-rigid human body target is decomposed into 25 joints tracked by an optical device which is Kinect in this paper. After the decomposition, the movement of each joint can be described as a space-time trajectory tracked by Kinect, providing a preliminary indication of the 3D position of each joint. With the indication, we divide every microwave/mmw image into 25 sub-images corresponding to the joints' position to form an image sequence of each joint. Subsequently, a segmental joint estimation method is proposed to estimate the accurate motion parameters of each joint from the image sequence. The decorrelation of the images in large angle observation can be significantly reduced by the segmental estimation which leads global optimized motion parameters. Consequently, the estimated parameters are taken into the effective movement compensation of each component. Meanwhile, the image refocusing method is presented to generate the well-focused image of each component. Finally, all the images of components are fused into one image of the whole target to reconstruct the 3D high-resolution image. Imaging simulations and measuring data experiments show that this method is efficient.
What problem does this paper attempt to address?