A Novel Image Dehazing and Assessment Method
This work addresses image quality issues for computer vision applications in hazy environments, but it appears incremental as it builds on existing dehazing methods with a modified approach and new assessment metrics.
The paper tackles image degradation in hazy conditions by proposing a method that estimates the transmission map using haze levels instead of airlight color, showing competitive results in qualitative and quantitative evaluations, and introduces two new metrics based on natural outdoor image statistics for assessing haze removal algorithms.
Images captured in hazy weather conditions often suffer from color contrast and color fidelity. This degradation is represented by transmission map which represents the amount of attenuation and airlight which represents the color of additive noise. In this paper, we have proposed a method to estimate the transmission map using haze levels instead of airlight color since there are some ambiguities in estimation of airlight. Qualitative and quantitative results of proposed method show competitiveness of the method given. In addition we have proposed two metrics which are based on statistics of natural outdoor images for assessment of haze removal algorithms.