Abstract
Dark Channel Prior (DCP) is a widely recognized traditional dehazing algorithm. However, it may fail in bright region and the brightness of the restored image is darker than hazy image. In this paper, we propose an effective method to optimize DCP. We build a multiple linear regression haze-removal model based on DCP atmospheric scattering model and train this model with RESIDE dataset, which aims to reduce the unexpected errors caused by the rough estimations of transmission map t(x) and atmospheric light A. The RESIDE dataset provides enough synthetic hazy images and their corresponding groundtruth images to train and test. We compare the performances of different dehazing algorithms in terms of two important full-reference metrics, the peak-signal-to-noise ratio (PSNR) as well as the structural similarity index measure (SSIM). The experiment results show that our model gets highest SSIM value and its PSNR value is also higher than most of state-of-the-art dehazing algorithms. Our results also overcome the weakness of DCP on real-world hazy images
Highlights
Images captured in outdoor scenes are usually degraded by haze, fog and smoke
By demonstrating our dehazing results on several groups of hazy images as well as comparing the peak-signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM) value, we show the better performance of our model over dark channel prior (DCP) and other dehazing algorithms
A benchmark dataset of both synthetic and real-world hazy images provided in [3] for dehazing problems are introduced to the community
Summary
Images captured in outdoor scenes are usually degraded by haze, fog and smoke. Suffering from poor visibility, reduced contrasts, fainted surfaces and color shift, hazy images will miss many details. The dark channel prior (DCP) [1] is widely recognized. The traditional dark channel prior has a rough estimation on the transmission t(x) and the Atmospheric Light A. This weakness is very obvious when the scene object is inherently similar to the air light over a large local region and no shadow is cast on it, the brightness of the restored image will be darker than real haze-free image. We propose an improved multiple linear regression haze-removal model to optimize the accuracy of estimation on the transmission t(x) and the Atmospheric Light A. The haze-removal model is trained with the training set of REalistic Single Image DEhazing (RESIDE) dataset [3].
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
