Multi-Iterative Superpixel Segmentation based on Local Brightness and Darkness Information

Zheng, Junbao; Zhang, Sizhe; Xu, Chenke

doi:10.1109/iccst53801.2021.00102

Cited by 1 publication

(1 citation statement)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Combined with the experi-mental results, we believe that the sudden increase in the brightness value of the center pixel was affected by the surrounding pixels. In order to find the correct measurement point of atmospheric illumination, this paper proposes the shading partition enhancement mechanism [25].…”

Section: The Shading Partition Enhancementmentioning

confidence: 99%

Single Image Dehazing Using Global Illumination Compensation

Zheng

Zhang

et al. 2022

Sensors

View full text Add to dashboard Cite

The existing dehazing algorithms hardly consider background interference in the process of estimating the atmospheric illumination value and transmittance, resulting in an unsatisfactory dehazing effect. In order to solve the problem, this paper proposes a novel global illumination compensation-based image-dehazing algorithm (GIC). The GIC method compensates for the intensity of light scattered when light passes through atmospheric particles such as fog. Firstly, the illumination compensation was accomplished in the CIELab color space using the shading partition enhancement mechanism. Secondly, the atmospheric illumination values and transmittance parameters of these enhanced images were computed to improve the performance of atmospheric-scattering models, in order to reduce the interference of background signals. Eventually, the dehazing result maps with reduced background interference were obtained with the computed atmospheric-scattering model. The dehazing experiments were carried out on the public data set, and the dehazing results of the foggy image were compared with cutting-edge dehazing algorithms. The experimental results illustrate that the proposed GIC algorithm shows enhanced consistency with the real-imaging situation in estimating atmospheric illumination and transmittance. Compared with established image-dehazing methods, the peak signal-to-noise ratio (PSNR) and the structural similarity (SSIM) metrics of the proposed GIC method increased by 3.25 and 0.084, respectively.

show abstract

Section: The Shading Partition Enhancementmentioning

confidence: 99%