VideoSAR is a land-imaging mode where the synthetic aperture radar (SAR) is operated in a spotlight configuration for an extended period of time. A sequence of images is continuously formed to a common Cartesian grid, while the radar is either flying towards, by or circling a target area. In general, VideoSAR imaging maintains antenna illumination on a target regardless of changes in squint angle, within theoretical and practical limitations. The video-like nature is a result of the imagery being produced using overlapped synthetic apertures such that the output frame rate can be commensurate with that of a video system. Enhanced exploitation of this product typically requires that the imagery be phase preserving, and that the data acquisition geometries be highly controlled for generation of secondary products such as coherence images (coherence maps) for change detection. One of the most accurate image formation algorithms for the formation of phase preserved imagery is the back-projection algorithm. Though computationally intensive, its attraction lies in its simplicity, the fact that the relative location of each pixel in the output imagery is precisely known and the complex imagery is phase preserved. The backprojection algorithm is presented in terms of its application to VideoSAR imaging. Mathematical decomposition techniques for improving the computational efficiency of the algorithm are reviewed. The application of change detection between pairs of VideoSAR images and `stacks' of VideoSAR images where the benefits of trading off spatial image filtering and temporal image filtering by averaging coherence in slow-time is also examined. Finally, computer topologies applicable to back projection are outlined from the perspectives of both the signal processing architecture and advances in massively parallel computing.

Chapter Contents:

• Abstract
• 4.1 Introduction
• 4.2 VideoSAR imaging
• 4.2.1 Image formation theory
• 4.2.2 Back-projection for VideoSAR image formation
• 4.2.3 Non-coherent integration of back-projected images
• 4.2.4 Image ambiguities
• 4.2.5 Autofocus considerations
• 4.3 Change detection
• 4.3.1 Background
• 4.3.2 Amplitude change detection
• 4.3.3 Coherent change detection
• 4.4 Real-time VideoSAR implementation
• 4.4.1 Challenges mapping core VideoSAR algorithms onto GPGPUs
• 4.4.2 System overview
• 4.4.3 High-speed data acquisition
• 4.4.4 High-performance computer
• 4.4.5 SAR image formation software
• 4.4.6 VideoSAR visualization
• 4.4.7 Image compression and scaling
• 4.4.8 VideoSAR GUI
• 4.5 Summary and outlook
• List of acronyms and abbreviations
• References

Inspec keywords: image filtering; radar imaging; synthetic aperture radar; video surveillance; radar antennas

Other keywords: temporal image filtering; common Cartesian grid; change detection; back-projection algorithm; overlapped synthetic apertures; real-time persistent surveillance; complex imagery; synthetic aperture radar; antenna illumination; mathematical decomposition; videoSAR imaging; spatial image filtering; phase preserved imagery; data acquisition geometry; computer topology; massively parallel computing; spotlight configuration; land-imaging mode; image formation; signal processing architecture

Subjects: Radar theory; Optical, image and video signal processing; Single antennas; Radar equipment, systems and applications