Information-Preserved Blending Method for Forward-Looking Sonar Mosaicing in Non-Ideal System Configuration

TL;DR

This work tackles the degradation of forward-looking sonar mosaics under non-ideal system configurations by proposing a blending framework that preserves information through a Global Variance Map (GVM) and a Long-Short Time Sliding Window (LST-SW). Offset estimation is performed via a Fourier–Mellin Transform to align frames in SE(2), followed by blending that excludes featureless background pixels, retaining informative content $ left O ight$ determined by the GVM. Local statistics are refined with per-pixel variance $v_{k,t}$ and scores $s_{k,t}$ computed using short and long time windows $(L_s,L_l)$, i.e., $v_{k,t,o}= abla$ the short-window sum, $v_{k,t,b}= abla$ the long-window sum, and $s_{k,t}=v_{k,t,o} e^{-v_{k,t,b}}$, to form the GVM $M_v$ and select informative pixels. Real-environment experiments demonstrate that the method preserves more texture and detail than plain averaging, enabling more reliable human inspection and potentially faster decision-making in underwater inspection tasks; adaptive and real-time implementations are suggested for future work.

Abstract

Forward-Looking Sonar (FLS) has started to gain attention in the field of near-bottom close-range underwater inspection because of its high resolution and high framerate features. Although Automatic Target Recognition (ATR) algorithms have been applied tentatively for object-searching tasks, human supervision is still indispensable, especially when involving critical areas. A clear FLS mosaic containing all suspicious information is in demand to help experts deal with tremendous perception data. However, previous work only considered that FLS is working in an ideal system configuration, which assumes an appropriate sonar imaging setup and the availability of accurate positioning data. Without those promises, the intra-frame and inter-frame artifacts will appear and degrade the quality of the final mosaic by making the information of interest invisible. In this paper, we propose a novel blending method for FLS mosaicing which can preserve interested information. A Long-Short Time Sliding Window (LST-SW) is designed to rectify the local statistics of raw sonar images. The statistics are then utilized to construct a Global Variance Map (GVM). The GVM helps to emphasize the useful information contained in images in the blending phase by classifying the informative and featureless pixels, thereby enhancing the quality of final mosaic. The method is verified using data collected in the real environment. The results show that our method can preserve more details in FLS mosaics for human inspection purposes in practice.

Figures (4)

• Figure 1: An open, box-shaped, lattice iron basket lying on the seabed imaged by FLS. (a) Strong reflection from the basket is overwhelmed by background. (b) A better result of another shot. The images are presented in raw (polar) format.
• Figure 2: Illustration of inter-frame artifacts. The brighter white square on the left side represents an angumentation. On the contrary, the fainter one on the right side represents a result of over-smooth.
• Figure 3: The Huangjinwan reservoir which located on the northwest coast of Zhoushan City, Zhejiang Province, China. Image from Google Earth.
• Figure 4: The comparison of the mosaics constructed by the plain average and the proposed method. The zoomed areas are framed by different colors.