HazeRD: an outdoor dataset for dehazing algorithms
- Citation Author(s):
- Submitted by:
- Yanfu Zhang
- Last updated:
- DOI:
- 10.21227/H2001Q
- Data Format:
2597 views
- Categories:
- Keywords:
Abstract
HazeRD is an outdoor scene dataset for benchmarking dehazing algorithms. HazeRD contains 10 different scenes based on the architectural biometrics project. For each scene, the ground RGB images, depth maps, and synthesized hazy images following the atmospheric optics are provided; the hazy images come with five different haze level using real life physical parameters. The main features of HazeRD to other dehazing datasets are: HazeRD focuses on outdoor scenes whereas other datasets provide indoor scenes; and, the synthesis is based on real life parameters.
Instructions:
The image files are organized as each scene with synthesized images under five different weather conditions. We also provide the matlab scripts for the generation of hazy images.
The files are named as follow:
hazy_img_generation.m
scene_{1~10}/
gt.png
depth.png
haze_{50,100,200,500,1000}.png
In reply to Good by FAN ZIR
Could not find the files.