Diego Henrique Diemmer Mariani

Diego Henrique Diemmer Mariani

Auto-generated profile, based on publicly available information

Diego Henrique Diemmer Mariani

VIZLAB Center of Excellence in Geoinformatics and Visual Computing, Graduate Program in Applied Computing (PPGCA), Universidade do Vale do Rio dos Sinos (UNISINOS), São Leopoldo, RS, Brazil

Research Focus & Metrics

Academic Career Length

1Year (2026-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
9 June 2026 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Verified
Journal Article
Open access
DMDiego Henrique Diemmer Mariani
Diego Henrique Diemmer Mariani
LSLucas Adams Seewald
Lucas Adams Seewald
Leandro ToniettoLeandro Tonietto
Leandro Tonietto
IBIsabel Borges
Isabel Borges
ASAndré Durante Spigolon
André Durante Spigolon
MVMAURICIO ROBERTO VERONEZ
MAURICIO ROBERTO VERONEZ
In the context of seismic interpretation, this work proposes a semi-automatic strategy for reconstructing spatially continuous seismic horizons from three-dimensional seismic volumes using limited manual input. The method is initialized from user-defined seeds placed on inline or crossline sections and combines horizon tracking with image-based segmentation, orthogonal propagation between seismic sections, contour-based interpolation, confidence-aware segmentation management, and mesh-based surface reconstruction. Horizon information is progressively propagated across orthogonal seismic sections by exploiting their geometric intersections, enabling controlled expansion from a small number of manually inserted seeds. The proposed workflow was evaluated on the Delft 3D seismic dataset and compared with the OpendTect automatic horizon-tracking workflow using the same initialization strategy. Quantitative evaluation was performed using overlap-based metrics such as Dice coefficient and Intersection over Union (IoU), and distance-based measures including Average Hausdorff Distance (AHD), Hausdorff Distance (HD), Mean Absolute Error (MAE), and Root Mean Squared Error (RMSE), using manually interpreted horizons as ground truth. The proposed method achieved average Dice scores of 0.826 and 0.832 for crossline and inline sections, respectively, while consistently reducing geometric deviation relative to OpendTect. Average RMSE was reduced from 9.909 to 2.149 in crossline sections and from 13.828 to 2.706 in inline sections. Visual analysis of the reconstructed surfaces also showed improved spatial continuity and fewer propagation gaps. The results indicate that the proposed method can reconstruct spatially coherent seismic horizons while maintaining geometric consistency under limited manual interaction.