Detection of Boulders in Side Scan Sonar Mosaics by a Neural Network

Abstract

Boulders provide ecologically important hard grounds in shelf seas, and form protected habitats under the European Habitats Directive. Boulders on the seafloor can usually be recognized in backscatter mosaics due to a characteristic pattern of high backscatter intensity followed by an acoustic shadow. The manual identification of boulders on mosaics is tedious and subjective, and thus could benefit from automation. In this study, we train an object detection framework, RetinaNet, based on a neural network backbone, ResNet, to detect boulders in backscatter mosaics derived from a sidescan-sonar operating at 384 kHz. A training dataset comprising 4617 boulders and 2005 negative examples similar to boulders was used to train RetinaNet. The trained model was applied to a test area located in the Kriegers Flak area (Baltic Sea), and the results compared to mosaic interpretation by expert analysis. Some misclassification of water column noise and boundaries of artificial plough marks occurs, but the results of the trained model are comparable to the human interpretation. While the trained model correctly identified a higher number of boulders, the human interpreter had an advantage at recognizing smaller objects comprising a bounding box of less than 7 × 7 pixels. Almost identical performance between the best model and expert analysis was found when classifying boulder density into three classes (0, 1-5, more than 5) over 10,000 m 2 areas, with the best performing model reaching an agreement with the human interpretation of 90%.

Limitations of Boulder Detection in Shallow Water Habitats Using High-Resolution Sidescan Sonar Images

“…Stone and boulder assemblages enhance small-scale spatial habitat variability. Their distribution and the potential settlement space they create is not quantitatively known [27]. The resolution of SSS images strongly affects the quantitative identification of stones and boulders.…”

“…This study demonstrates the practical application of SSS in identifying vegetated stones and boulders in very shallow waters. The suitability of high-resolution acoustic imaging techniques has previously been reported in areas with water depths >10 m [13,16,18,27]. Satellite remote sensing and aerial photographs offer another possibility for characterizing shallow water habitats, but high costs and limited resolutions are disadvantages for these methods [23].…”

“…The identification of objects using multibeam echo sounders (MBES) has been shown for, e.g., mine detection [20,26]. Area-wide habitat mapping is often performed by using acoustic frequencies ranging from 300 to 500 kHz, survey speeds vary between 2-4 m/s and across-track ranges between 50 m and 180 m [13,16,18,27]. Those survey setups generate SSS image resolutions ranging from 0.25 m/pixel to 1 m/pixel, thus are insufficient to identify individual stones or boulders, especially in areas with coarse, mixed sediments [16].…”

“…Those survey setups generate SSS image resolutions ranging from 0.25 m/pixel to 1 m/pixel, thus are insufficient to identify individual stones or boulders, especially in areas with coarse, mixed sediments [16]. Despite recent advances in methods for the automated detection of individual stones and boulders [27,28] methodological constraints of survey settings (acoustic frequencies, horizontal resolution and range, survey speeds, etc.) remain undiscussed.…”

Automatic Detection of Trawl-Marks in Sidescan Sonar Images through Spatial Domain Filtering, Employing Haar-Like Features and Morphological Operations

“…Malik and Mayer [44] noted that video images depict the TMs with significantly less accuracy than SSS, while Tang et al [45] analyzed the trawling pattern in video images with a neural classifier. Despite the fact that there has recently been strong interest in deep learning methods for geological seafloor mapping [46,47] there are no studies using machine learning methods for TM recognition in SSS images. To the best of our knowledge, up to today there is no TM detection algorithm in the literature applied to a large SSS mosaic that could detect and quantify TMs for their environmental impact monitoring use.…”