Title : Single Image Dehazing and Defogging for the estimation of optimal transmission map using restoration techniques
Abstract:
Bad weather has an adverse impact on outdoor vision systems and also increases the possibilities of traffic crashes and road accidents. Outdoor images captured in bad weather conditions such as haze and fog are degraded due to the scattering of atmospheric particles. The decreased visibility significantly impacts on driving assistance systems developed for automatic vehicle. So, clear visibility is very essential for the outdoor computer vision systems. Image deweathering is one of the ill posed problems since the transmission depth estimation is very uncertain. Therefore, the most accurate estimation of transmission depth is indispensable. Traditional state-of-the-art restoration techniques frequently generate halo artifacts or reduce the contrast of the observed object in the restored images due to unequal transmission depth distribution. The most difficult aspect of previous state-of-the-art techniques is accurately and effectively estimating the transmission depth from object to object in a short period of time. The restoration-based approaches are used to estimate the refinement of the transmission map to recover scene radiance. The following performance metrics are used for quantitative analysis of the experimental results: contrast gain, blind contrast assessment (e, r), peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), and image visibility measurement (IVM).
What will audience learn from your presentation?
? Scattering phenomenon for atmospheric particles
? Image degradation model to understand the formation of image in presence of bad weather
conditions
? Image processing algorithms and different approaches would help to researcher or academician to
expand their research in different domains. The restoration based methods provides a suitable
solutions for real time implementation. The solution will reduce the complexity of processing time
and restore the clear and visible image