HiCo-GS: Hierarchical Context Aggregation and Geometric Consistency for Octree Gaussian Splatting

2026-08-14

AI Summary

This study addresses color drift and texture over-smoothing caused by cross-layer feature isolation in octree-based Gaussian Splatting. We propose HiCo-GS, a novel framework that employs a cross-layer context aggregation module to enable bidirectional prior injection between hierarchical levels. Furthermore, it incorporates depth-normal geometric consistency regularization to effectively suppress floating artifacts. In urban-scale novel view synthesis tasks, HiCo-GS overcomes the bottleneck of hierarchical feature isolation, achieving state-of-the-art rendering quality and high-fidelity geometric reconstruction across multiple benchmark datasets. Consequently, this method significantly enhances both the accuracy and consistency of scene representation, demonstrating superior performance in complex large-scale environments compared to existing approaches.

Abstract

Octree-based anchor Gaussian Splatting has emerged as a scalable representation for city-scale novel view synthesis, where multi-level anchors adaptively capture scene content from coarse building structures to fine architectural details. However, we identify a fundamental limitation in existing methods: cross-level feature isolation, where each level's anchor features are optimized independently with no inter-level communication, causing color drift on building facades and over-smoothing in textured regions. We present HiCo-GS, a high-fidelity reconstruction framework with two complementary modules. Cross-Level Context Aggregation (CLCA) enables bidirectional hierarchical prior injection by leveraging the octree's spatial containment structure to aggregate per-level context vectors into parent-self-child triplets, fused via a lightweight MLP with residual connection. Coarse-level structural priors flow down to inform fine-level anchors, while fine-level detail statistics feed back to prevent over-smoothing, at negligible computational overhead. Depth-Normal Geometric Consistency (DNGC) regularization enforces agreement between rendered normals and depth-derived normals through an alpha-weighted consistency loss, complemented by edge-aware smoothness losses with progressive warmup that exploit the strong planar priors ubiquitous in urban geometry to suppress floating artifacts. We further introduce the China-Pagoda dataset comprising 8 ancient Chinese pagodas with over 1,200 images each, featuring dense ornamental carvings, curved multi-layer eaves, and repetitive fine-grained textures. Extensive experiments on Mill19, UrbanScene3D, MatrixCity, and China-Pagoda demonstrate that HiCo-GS achieves state-of-the-art rendering quality and substantially cleaner geometry across real-world and synthetic urban benchmarks.Code: https://github.com/WZ-CS/HiCo-GS.

Problem

• Octree Gaussian Splatting
• Cross-level feature isolation
• Novel view synthesis
• Color drift
• Over-smoothing

Innovation

• Cross-Level Context Aggregation
• Depth-Normal Geometric Consistency
• Octree Gaussian Splatting
• Hierarchical Prior Injection
• China-Pagoda Dataset