Research And System Iplementation On Image Vedio-surveillance-orientated Dehazing Method

The outdoor surveillance system is more sensitive to extreme weather,such as fog and haze.In the fog and haze weather conditions,the outdoor surveillance performance is severely affected,which leads to the video quality severely degraded.Therefore,in order to enhance the applicability and stability of outdoor video surveillance,it is significant to study the fog and haze image restoration techniques.The main works and contribution of this thesis are summarized as bellows:1.In order to avoid the halo phenomenon of the dark channel prior algorithm,a single image dehazing method based on luminance image is proposed.First,the method of estimating transmission map in the dark channel prior is improved.The minimum channel is replaced by the luminance component of the YIQ color model and the median filtering takes the place of the minimum filtering,which improves the computational efficiency of the dark channel prior algorithm and inhibits the halo phenomenon of the dark channel prior algorithm;Next,a higher efficiency of the four binary tree algorithm is used to obtain the atmospheric light value;Finally,the experiment is conducted on various kinds of haze images.The experimental results show that the processing speed and dehazing effects of this method are superior to other kinds of classical algorithms.The single image dehazing method in this paper is also suitable for the vedio dehazing.2.The dehazing method of oriented vedio surveillance is proposed.First,according to the characteristic of surveillance vedios,the universal background image is extracted by background modeling method of mean filtering;Next,the universal background is used to estimate the transmission map and the atmospheric light value;Finally,the atmospheric scattering model is used to recover each frame and the clear vedio is obtained.The experimental results show that the dehazing method of oriented surveillance vedio is close to the effect of real time dehazing.3.On the basis of the above work,designing and implementing a single image dehazing system on MATLAB7.9.0 platform and a vedio dehazing system on Visual Studio 2010 platform.These two systems inherit some classic algorithms.