A Comprehensive Picking Method for Main and Detailed Horizon in 3D Sub-Bottom Profilers

Objectives: Three-dimensional sub-bottom profilers (3D SBP), due to its ability to reflect stratigraphic structural information in three-dimensional space, plays a significant role in marine scientific research and engineering applications. However, existing methods for picking horizons from 3D SBP typically rely on single features, such as echo intensity characteristics or structural attributes, which makes it difficult to simultaneously achieve the picking of both main and detailed horizons. To overcome this limitation, this paper proposes an integrated picking method for main and detailed horizons in 3D SBP based on an active contour model. Methods: The proposed method adopts an active contour model framework. First, main horizon structures are enhanced via horizon enhancement filtering. Conventional median and mean filtering methods often fail to account for local structural characteristics of the image, leading to image distortion and quality degradation. To overcome this, based on the directional information derived from the enhancement filtering, a detail preserving filtering approach is proposed to retain the fine structural features of horizons. Subsequently, the Region Scalable Fitting (RSF) active contour model is introduced. A main horizon constraint term is constructed using the enhancement filtering results, while a detailed horizon constraint term is formulated based on the direction preserving filtering results. Finally, the active contour model is optimized through level set evolution to achieve integrated picking of 3D main and detailed horizons under multiple constraints. Results: Experiments were conducted on multiple datasets with accuracy analysis. The results demonstrate that for 3D SBP data with relatively simple horizon structures and for data from complex regions, the comprehensive evaluation metric F-measure of the proposed method reaches 95.8% and 85.4%, respectively, indicating robust performance. Compared with the traditional RSF method and enhancement filtering based picking approaches, the proposed method shows improved accuracy in horizon picking, effectively integrating both main and detailed structural information. Conclusions: By effectively incorporating multiple feature constraints into the active contour model framework, the proposed method enables simultaneous and accurate picking of main and detailed horizons from 3D SBP data. This integrated approach provides a more comprehensive and reliable solution for interpreting complex shallow submarine structures, demonstrating great potential for high-precision marine geological surveys and sub-bottom profiling analysis. It also holds promise for applications such as sediment classification and geological structure analysis. It is worth noting that the RSF model was selected in this work to achieve integrated picking of main and detailed horizons; however, other active contour models capable of handling more complex topological changes and processing images with intensity inhomogeneity may also serve as viable alternatives.