2025_programme: An Improved Backprojection Autofocus for Synthetic Aperture Sonar

• Day: June 19, Thursday
    Location / Time: D. CHLOE at 11:00-11:20
• Last minutes changes: Cancelled
• Session: 11. New methods and theories in underwater acoustic imaging
  Organiser(s): Pengfei Zhang, Peng Wang
  Chairperson(s): Pengfei Zhang, Peng Wang
• Lecture: An Improved Backprojection Autofocus for Synthetic Aperture Sonar
  Paper ID: 2188
  Author(s): Shiping Chen, Lulu Ren, Pengfei Zhang, Peng Wang, Cheng Chi, Jiyuan Liu, Yu Li, Haining Huang
  Presenter: Shiping Chen
  Abstract: The autofocus methods can effectively compensate the multi-source phase errors in synthetic aperture radar (SAR) and synthetic aperture sonar (SAS). The backprojection (BP) autofocus has attracted attention due to its flexibility and fewer limitations. The BP autofocus algorithms based on image sharpness or entropy have been widely applied in SAR, yet their application in SAS has rarely been reported. However, the BP autofocus based on maximizing sharpness (or minimizing entropy) bears a significant computation burden due to the necessity of conducting an optimal iterative search for phase errors and involving a vast number of pixels in the computation process. In this paper, we propose an improved BP autofocus algorithm to reduce the computation burden of the conventional algorithms. The proposed algorithm includes two acceleration strategies. Firstly, Anderson acceleration, a technique designed to accelerate iterative search processes, is employed in the iterations of the autofocus algorithm, with the objective of reducing time cost by minimizing the number of iterations required. Secondly, we have discovered that target pixels (those with higher energy) are sparse in SAS images, yet they exert a decisive influence on the statistical characteristics of the images. By extracting these target pixels and employing only them in the iterative search process of the BP autofocus, we can significantly reduce the number of pixels involved in the computation, thereby alleviating the computation burden of the algorithm. Data collected in the field were utilized to test the proposed algorithm. The results demonstrate that, compared to two conventional algorithms, the proposed algorithm achieved comparable image quality while reducing the time cost by 6.88 and 111.33 times, respectively. The reduction in time cost achieved by the proposed method may facilitate the application of BP autofocus in real-time systems.
• Corresponding author: Dr Shiping Chen
  Affiliation: Institute of Acoustics,Chinese Academy of Sciences
  Country: China