The Stage-I GAN suppresses the multipath ghosts based on the original input images, while side/grating lobe artifacts are intentionally preserved in the stage-I GAN as ad...
Abstract:
For multi-input multi-output (MIMO) through-wall radar imaging (TWRI), multipath ghosts and side/grating lobe artifacts degrade the imaging quality of the obscured target...Show MoreMetadata
Abstract:
For multi-input multi-output (MIMO) through-wall radar imaging (TWRI), multipath ghosts and side/grating lobe artifacts degrade the imaging quality of the obscured targets inside an enclosed building, therein hindering target detection. In this paper, an approach based on two stacked generative adversarial nets (GAN) is proposed to achieve multipath and side/grating lobe suppression with regard to MIMO TWRI. Specifically, the Stage-I GAN constructs a spatial structure mapping from the original input images to the Stage-I GAN output images with the suppressed multipath ghosts. However, the side/grating lobe artifacts are intentionally preserved in the stage-I GAN as additional constraint information to prevent uncontrollable over-fitting. Then, the Stage-II GAN takes the output images of Stage-I GAN as input to suppress the side/grating lobe artifacts. Extensive electromagnetic simulations and comparisons demonstrate that the proposed approach achieves better suppression of multipath ghosts and side/grating lobe artifacts and other significant superiorities, including priori wall information not being required, the preservation of weak targets, and robustness for different array deployments and building layouts.
The Stage-I GAN suppresses the multipath ghosts based on the original input images, while side/grating lobe artifacts are intentionally preserved in the stage-I GAN as ad...
Published in: IEEE Access ( Volume: 7)
Funding Agency:
Keywords assist with retrieval of results and provide a means to discovering other relevant content. Learn more.
- IEEE Keywords
- Index Terms
- Radar Images ,
- Multi-input Multi-output ,
- Generative Adversarial Nets ,
- Through-wall Radar Imaging ,
- Through-wall Radar ,
- Input Image ,
- Spatial Structure ,
- Output Image ,
- Electromagnetic Simulation ,
- Weak Targets ,
- Poor Performance ,
- Objective Function ,
- Target Image ,
- Transmission Coefficient ,
- Wall Surface ,
- Sidelobe ,
- Finite-difference Time-domain ,
- Ground Truth Image ,
- Comparison Of Different Methods ,
- Relative Dielectric Constant ,
- Front Wall ,
- Coherence Factor ,
- Back-projection Algorithm ,
- Target Coordinates ,
- Radar Resolution ,
- CPU Processor ,
- Alternative Training ,
- Green Rectangle ,
- Back Wall ,
- Adverse Relationship
- Author Keywords
Keywords assist with retrieval of results and provide a means to discovering other relevant content. Learn more.
- IEEE Keywords
- Index Terms
- Radar Images ,
- Multi-input Multi-output ,
- Generative Adversarial Nets ,
- Through-wall Radar Imaging ,
- Through-wall Radar ,
- Input Image ,
- Spatial Structure ,
- Output Image ,
- Electromagnetic Simulation ,
- Weak Targets ,
- Poor Performance ,
- Objective Function ,
- Target Image ,
- Transmission Coefficient ,
- Wall Surface ,
- Sidelobe ,
- Finite-difference Time-domain ,
- Ground Truth Image ,
- Comparison Of Different Methods ,
- Relative Dielectric Constant ,
- Front Wall ,
- Coherence Factor ,
- Back-projection Algorithm ,
- Target Coordinates ,
- Radar Resolution ,
- CPU Processor ,
- Alternative Training ,
- Green Rectangle ,
- Back Wall ,
- Adverse Relationship
- Author Keywords