A Spectral Hazy Image Database

Abstract

We introduce a new database to promote visibility enhancement techniques intended for spectral image dehazing. SHIA (Spectral Hazy Image database for Assessment) is composed of two real indoor scenes M1 and M2 of 10 levels of fog each and their corresponding hazefree (ground-truth) images, taken in the visible and the near infrared ranges every 10nm starting from 450 to 1000nm. Thus, the number of images that form SHIA is 1540 with a size of 1312 × 1082 pixels. The hazy images and the haze-free images are captured under the same illumination conditions. Three of the well-known dehazing image methods belonging to different categories were adjusted and applied on the spectral hazy images. This study confirms once again a strong dependency between dehazing methods and fog densities. It urges the design of spectral-based image dehazing able to handle simultaneously the accurate estimation of the parameters of the visibility degradation model and the limitation of artifacts and post-dehazing noise. The database can be downloaded freely at http://chic.u-bourgogne.fr.

Single Image Dehazing Algorithm Analysis with Hyperspectral Images in the Visible Range

“…SHIA consists of two real indoor scenes, M1 and M2, under ten different levels of fog each (level 0 corresponds to the maximum level of fog, whereas level 9 to the minimum), generated with a fog machine, and their corresponding fog-free (ground truth) image, captured in the visible and the near infrared (450 to 1000 nm with a spectral resolution of 10 nm). According to El Khoury et al [ 17 ], the generated fog in the experiment, composed of water droplets, had radius values similar to real atmospheric fog.…”

“…We use, here, images from the spectral hazy image database for assessment (SHIA) [ 17 ]. SHIA consists of two real indoor scenes, M1 and M2, under ten different levels of fog each (level 0 corresponds to the maximum level of fog, whereas level 9 to the minimum), generated with a fog machine, and their corresponding fog-free (ground truth) image, captured in the visible and the near infrared (450 to 1000 nm with a spectral resolution of 10 nm).…”

“…However, despite their potential, they have not yet been used in dehazing methods. Recently, a spectral hazy image database for assessment (SHIA) has become available [ 17 ], which was inspired by a previous color image database (CHIC) from the same authors [ 18 ]. SHIA [ 17 ] includes a set of spectral images with various levels of haze/fog generated with a fog machine, with the corresponding fog-free spectral image which serves as the reference image (ground truth).…”

“…Recently, a spectral hazy image database for assessment (SHIA) has become available [ 17 ], which was inspired by a previous color image database (CHIC) from the same authors [ 18 ]. SHIA [ 17 ] includes a set of spectral images with various levels of haze/fog generated with a fog machine, with the corresponding fog-free spectral image which serves as the reference image (ground truth). The difference between haze and fog is based on the radius size of the water particles (less than 1 micron corresponds to haze) and the concentration (more than 100 particles/cm 3 is considered as fog) [ 19 ].…”

“…Do metrics for image quality assessment correlate with the perceptual judgment of human observers? El Khoury et al [ 17 ] also tackled some of these questions but, here, we have used different dehazing algorithms, including the recent Berman method [ 23 ], a more comprehensive set of metrics, and we incorporate a psychophysical experiment that was not considered in the El Khoury et al [ 17 ] paper.…”