Improved 3D scene recovery from hazy images with new Gaussian splatting method
Tri-DehazeGS: Scene--Medium Decoupled Gaussian Splatting with Transmittance-Aware Optimization
Computer Vision and Pattern Recognition
Summary
Haze in photos can make it hard to see what’s really in a 3D scene, because it blurs and dims objects. The authors developed a new technique called Tri-DehazeGS that separates the clean objects in a scene from the haze effects when reconstructing the scene in 3D. Their method models the haze separately and adjusts how the computer learns from hazy areas, helping it build clearer 3D views from multiple hazy images. Tests with real and simulated hazy pictures showed that their approach produces better clean images from new viewpoints.
What this means in practice
- •For computer vision developers: Create clearer 3D models from multiple images taken in foggy or hazy conditions by separating scene details from haze effects.
- •For augmented reality engineers: Enhance visual quality of AR scenes generated outdoors in hazy weather by improving the reconstruction of clean scene radiance.
Authors
Kui Jiang, Yang Gu, Jiacheng Liu, Shiyu Liu, Youyu Chen, Hui Liu
Abstract
Recovering clean 3D scenes from hazy multi-view images is challenging because haze attenuates scene radiance and introduces atmospheric scattering. Recent scattering-aware Gaussian Splatting methods introduce physical haze models into reconstruction, but they often apply degradation in image space or bind medium-related variables to Gaussian primitives, which can entangle clean scene radiance with atmospheric effects. Moreover, low-transmittance regions provide weakened supervision for Gaussian optimization, causing distant or dense-haze areas to be under-reconstructed. We argue that clean reconstruction under haze requires both scene--medium disentanglement and transmittance-aware optimization rebalancing. To this end, we propose Tri-DehazeGS, a scene--medium decoupled Gaussian Splatting framework. It represents the clean scene with Gaussian primitives, models the participating medium using an independent view-shared tri-plane field, and composes hazy observations through a physical scattering model. We further introduce Medium-Decoupled Transmittance Gradient Compensation (MD-TGC), which compensates haze-suppressed gradients after medium freezing without altering forward rendering. Experiments on real and synthetic haze benchmarks show that Tri-DehazeGS improves clean novel-view reconstruction. Code is available at https://github.com/aptx46/Tri-DehazeGS.