Shadow-Background-Noise 3D Spatial Decomposition Using Sparse Low-Rank Gaussian Properties for Video-SAR Moving Target Shadow Enhancement
This addresses a domain-specific problem in video-SAR imaging for military or surveillance applications, with incremental improvements in existing methods.
The paper tackles the problem of poor detection-tracking accuracy for moving target shadows in video-SAR images, which are interfered by backgrounds and noises, by proposing a 3D spatial decomposition model that enhances shadows and boosts detection accuracy of models like Faster R-CNN and tracking accuracy of models like TransTrack.
Moving target shadows among video synthetic aperture radar (Video-SAR) images are always interfered by low scattering backgrounds and cluttered noises, causing poor detec-tion-tracking accuracy. Thus, a shadow-background-noise 3D spatial decomposition (SBN-3D-SD) model is proposed to enhance shadows for higher detection-tracking accuracy. It leverages the sparse property of shadows, the low-rank property of back-grounds, and the Gaussian property of noises to perform 3D spatial three-decomposition. It separates shadows from back-grounds and noises by the alternating direction method of multi-pliers (ADMM). Results on the Sandia National Laboratories (SNL) data verify its effectiveness. It boosts the shadow saliency from the qualitative and quantitative evaluation. It boosts the shadow detection accuracy of Faster R-CNN, RetinaNet and YOLOv3. It also boosts the shadow tracking accuracy of TransTrack, FairMOT and ByteTrack.