CNN‐based multiple‐input multiple‐output radar image enhancement method. Issue 20 (13th September 2019)

• Record Type: Journal Article
• Authors: Dai, Yongpeng; Jin, Tian; Song, Yongping; Du, Hao; Zhao, Dizhi
• Is Part Of: Journal of engineering. Volume 2019:Issue 20(2019)
• Page Start: 6840
• Page End: 6844
• Publication Date: 2019-09-13
• Subjects: radar cross‐sections -- radar imaging -- MIMO radar -- image resolution -- radar resolution -- image enhancement -- radar computing -- convolutional neural nets; complex valued two‐dimensional low‐resolution radar image -- CNN‐based multiple‐input multiple‐output radar image enhancement method -- coherence factor method -- grating lobe -- sidelobe -- radar‐cross‐section distribution image -- four‐layer convolutional neural network; Engineering -- Periodicals; Engineering; Electronic journals; Periodicals; 620.005
• DOI: 10.1049/joe.2019.0543
• Languages: English
• ISSNs: 2051-3305

Abstract

Abstract : In this study, the convolutional neural network (CNN) is utilised to enhance the quality of radar images. First, a four‐layer convolutional neural network is trained. The input of it is a complex valued two‐dimensional low‐resolution radar image, and the output is the radar‐cross‐section distribution image. After processed by the proposed network, the sidelobe and grating lobe in the radar image are suppressed, the main lobe of the target is sharpened. Comparing to the commonly used coherence factor method, the proposed method can enhance the image while maintaining the amplitude scaling relation between targets. The feasibility of the proposed method is testified by both simulated and experimental results.