Physics-based Image De-hazing Algorithms Research

The ultimate goal of computer vision is to understand and explain the rich three-dimensional world. With the development of basic science (cognitive neuroscience and visual computing technologies) and with the growing of computer cost performance, the ap-plications of computer vision become more and more extensive, such as security detection, industrial measurement and monitoring, automatic navigation,monitoring without anyone and so on. But the current visual system in the design cannot do with the disturbance of the ad-verse weather. So it can only be used in fine weather conditions. However, for any visual sys-tems applied in outdoor, they inevitably face all kinds of weather conditions, such as haze, fog, rain, snow and so on. A robust outdoor vision system must have the ability to work well under any weather conditions. This is an issue of great practical significance.The restoration of fog-degraded images is a very challenging problem. Recently, the re-search of fog-degraded image restoration is still on the stage of starting. There has been an increased interest on this problem. We concentrated image restoration techniques based on atmospheric scattering model in this paper.This dissertation is based on atmospheric scattering model and dark channel prior. On the basis of analyzing the image degradation mechanisms we research the solution of hazy image clearness. I propose a novel fast defogging method based on the atmospheric scattering model and dark channel prior theory. The white balance is performed and the atmospheric scattering model is simplified prior to visibility restoration.In the inference process of the atmospheric veil, the coarser estimate is refined using a fast median filtering approach that preserves edges. Finally, the scene albedo is recovered by inverting this simplified model. The complexity of the proposed method is only a linear function of the number of input image pixels and this allows a very fast implementation. To a certain extent it solves the common problem in hazy image restoration-high complexity.A large number of results are presented, and the no-reference quality assessment method for de-hazed image is involved. Results on a variety of outdoor foggy images demonstrate that the proposed method achieves good restoration for contrast and color fidelity, resulting in a great improvement in image visibility which fully explain this de-hazing technology rea- sonable and effective. In the end, we summarize all researches in this paper, and give the di-rection of following research in future.