DerainNeRF: 3D Scene Estimation with Adhesive Waterdrop Removal

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Li, Yunhao, Wu, Jing, Zhao, Lingzhe, Liu, Peidong
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917625386237952
author Li, Yunhao
Wu, Jing
Zhao, Lingzhe
Liu, Peidong
author_facet Li, Yunhao
Wu, Jing
Zhao, Lingzhe
Liu, Peidong
contents When capturing images through the glass during rainy or snowy weather conditions, the resulting images often contain waterdrops adhered on the glass surface, and these waterdrops significantly degrade the image quality and performance of many computer vision algorithms. To tackle these limitations, we propose a method to reconstruct the clear 3D scene implicitly from multi-view images degraded by waterdrops. Our method exploits an attention network to predict the location of waterdrops and then train a Neural Radiance Fields to recover the 3D scene implicitly. By leveraging the strong scene representation capabilities of NeRF, our method can render high-quality novel-view images with waterdrops removed. Extensive experimental results on both synthetic and real datasets show that our method is able to generate clear 3D scenes and outperforms existing state-of-the-art (SOTA) image adhesive waterdrop removal methods.
format Preprint
id arxiv_https___arxiv_org_abs_2403_20013
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle DerainNeRF: 3D Scene Estimation with Adhesive Waterdrop Removal
Li, Yunhao
Wu, Jing
Zhao, Lingzhe
Liu, Peidong
Computer Vision and Pattern Recognition
When capturing images through the glass during rainy or snowy weather conditions, the resulting images often contain waterdrops adhered on the glass surface, and these waterdrops significantly degrade the image quality and performance of many computer vision algorithms. To tackle these limitations, we propose a method to reconstruct the clear 3D scene implicitly from multi-view images degraded by waterdrops. Our method exploits an attention network to predict the location of waterdrops and then train a Neural Radiance Fields to recover the 3D scene implicitly. By leveraging the strong scene representation capabilities of NeRF, our method can render high-quality novel-view images with waterdrops removed. Extensive experimental results on both synthetic and real datasets show that our method is able to generate clear 3D scenes and outperforms existing state-of-the-art (SOTA) image adhesive waterdrop removal methods.
title DerainNeRF: 3D Scene Estimation with Adhesive Waterdrop Removal
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2403.20013