Automated change detection using Synthetic Aperture Sonar imagery

Abstract

When resurveying a seafloor area of interest during change detection operations, an automated method to match found bottom objects with objects detected in a previous survey allows the surveyor to quickly sort new objects from old. The change detection system developed at the Naval Research Laboratory contains modules for automatic object detection, feature matching using shadow outlining, scene matching using control-point matching, and visualization capabilities. This system was developed for sidescan sonar surveys using instrumentation such as the high-frequency Marine Sonic Technology sidescan sonar. In this paper, the authors describe modifications to the sidescan-based system required to perform change detection using Synthetic Aperture Sonar (SAS) bottom imagery.

Synthetic Aperture Sonar low frequency vs. high frequency automatic contact generation

“…However, in the field, the combination of not knowing a priori where objects are, the low grazing angles involved, and the high prevalence of clutter on the order of the signal makes it highly unlikely that buried objects would be detected using this system and current signal processing (e.g., figure 2). [1] was applied to the high-and low-frequency imagery collected on the first day of the survey. Table I shows the visual vs automatic clutter detection.…”

“…Up to now we have based our efforts on Side Scan Sonar imagery [,23,4]. In this paper and the companion paper [1] we wish to preliminarily examine how our detection tools translate into efforts based on Synthetic Aperture Sonar (SAS). Specifically, we wish to compare or combine the Low and High Frequency SAS available on Coastal Systems' Small Synthetic Aperture Minehunter (SSAM) Autonomous Underwater Vehicle (AUV) [5,6].…”

Studying Deep Learning Based Co-Registration for Incoherent Change Detection

“…In general, change detection algorithms are divided into image based and symbolic based methods 1 . Image-based change detection directly compares two images while in symbolic-based change detection objects are detected in both images separately, georeferenced and then compared 2,3,4 . If the images are captured with a sidescan sonar the conventional image based approach is an incoherent change detection in which two intensity images are first co-registered, i.e., aligned with respect to each other 5,6 .…”

Markov-based approaches for ternary change detection between two high resolution synthetic aperture sonar tracks

“…Incoherent change detection methods only use the amplitude images (amplitude of the backscattered energy) to find changes. While symbolic or object-based approaches rely on the extraction of features of interest (ATR detection [15], [16], echo-shadow pairs [17], [18]) in both images before matching them by considering their local arrangement, imagebased methods [7], [19] only consider pixels intensity to detect potential changes by means of the computation of a difference image.…”