ABSTRACT
This chapter considers sparsity-driven moving target indication. A radar imaging system can recognize an animate target indoors by a change in target position. The course of that change can be continuously or nonconsecutively monitored by the radar. The target location over time can be guided by a particular model or characterized by a specific motion profile. In this case, using a sequence of radar pulses, target velocity and its derivatives, including acceleration, can be obtained by estimating the corresponding Doppler frequency and higher-order motion parameters. Alternatively, the radar may illuminate the scene at particular time instants to capture the target at different positions. Detection of target position variations by subtraction of data or images at different time instants is the foundation of change detection where details of motion profile are difficult to discern and become irrelevant to motion indication. These instants can be widely separated, several pulses apart, or consecutive to detect short human movements. Whether motion modeling and characterization is performed or change detection is applied, we show in this chapter that the targets, maintaining or migrating their range, can be localized with significantly reduced amount of data due to sparseness in space and velocity.
