For access to this article, please select a purchase option:
IET members benefit from discounts to all IET publications and free access to E&T Magazine. If you are an IET member, log in to your account and the discounts will automatically be applied.
Your recommendation has been sent to your librarian.
VideoSAR generates image sequence continuously at a high frame rate, which provides great advantages for moving target detection. A key technology in the field of VideoSAR is moving targets detection. Most of the existing moving target detection methods for VideoSAR image sequence are based on shadows. However, the target shadows are difficult to obtain when the sensitivity of the radar system is not high enough. Instead, we detect moving target in VideoSAR frame via the bright lines formed by Doppler shift based on deep CNN (convolutional neural networks). In order to improve the CNN detection accuracy, the RPCA (Robust Principal Component Analysis) is adopted in this paper to separate the foreground image of VideoSAR. After the foreground images are added to the dataset for training, the experimental result verifies that the problems of missed detection, false detection and inaccurate target positioning are significantly improved and the mAP (mean Average Precision) of YOLOv3 for VideoSAR moving target has been particularly increased from 70.99% to 81.69%.
Inspec keywords: image sequences; feature extraction; Doppler shift; principal component analysis; image motion analysis; video signal processing; radar imaging; object detection
Subjects: Principal component analysis; Video signal processing; Neural nets; Radar equipment, systems and applications; Computer vision and image processing techniques; Machine learning (artificial intelligence); Optical, image and video signal processing; Principal component analysis