Answers are grounded in this paper and cite their sources.
The precise monitoring of spatial fish distribution in offshore aquaculture was critical for the optimization of feeding efficiency and facility design, yet traditional methods were incapable of continuous three-dimensional tracking, whereas existing acoustic technique face challenges in the reconstruction of quantifiable spatial coordinates from sonar images. In this study, we proposed an innovative sonar image-driven 3D reconstruction framework. Using omnidirectional high-frequency sonar, spiral-scanned data were collected on the spatial distribution of Trachinotus ovatus in the “Dehai No. 1” offshore aquaculture cage, which had a radius of approximately 12 m, a depth of approximately 6 m, and contained about 4000 fish. Through annular zone segmentation using 59 rings, adaptive binarisation with artefact removal of netting and cage-frame echoes, and coordinate transformation via polar mapping based on posture-specific pixel-area calibration, two-dimensional sonar echoes were converted into probabilistic three-dimensional fish positions. Vertically stratified aggregation was observed, with the highest distribution density occurring in the mid-lower water layers from 4 to 6 m and peaking at approximately 5.5 m. Horizontally, the fish were showed as a consistent pattern of central aggregation within 6 m radius and avoidance of the peripheral cage area, likely owing to structural compression and edge disturbance. The proposed method enables non-invasive quantitative mapping of cage-scale dynamics, showing promise for intelligent feeding control and welfare management despite limitations of near-field occlusion.
No takes yet. Share an insight, caveat, or question.
Li et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: