Fast Haze Removal Research For Video Image Based On Dark Channel Prior

With the development of computer technology and machine vision technology,computer vision monitoring equipment has been widely used in panoramic monitoring,aerospace remote sensing and military field.Haze removal is important in target detection,tracking and recognition based on image characteristics.Therefore,it is of great importance to improve the quality of image by dehazing algorithm.For the project and research needs,this paper propose a set of video image fast dehazing algorithm based on the dark channel prior.Aimed at the characteristics of camera on ‘The Cloud' and real-time requirements of this project,the haze removal algorithm is proposed based on dark channel prior.To implement the real-time requirements by using guiding filtering,image sampling technology and optimization algorithm based on the expansion of the dark spot.The adaptive transmission rate optimization algorithm and night time haze removal method are proposed to solve the defect of dark channel prior.At last,we propose the color cast detection and correction algorithm for the unnatural image.In this paper,the atmospheric scattering model and imaging model are introduced,and the basic principle of the dark channel prior is studied.Then,through research and analysis of the principle,the dehaze algorithm based on the dark channel prior principle is established.Aiming at the low efficiency of the soft matting algorithm,the guide filter is proposed to refine the calculating speed.On the premise of not reducing the effect of image dehazing,the optimization algorithm for the under sampling and dark point expansion of the dark channel image is used to meet the real-time requirement.And the optimization methods above are combined with the frame difference method to realize the fast haze removing of video image.Adaptive transmission rate optimization algorithm is proposed to the image which includes much sky area and not suit to the dark channel prior principle.And the night time imaging model is put forward in order to meet the night foggy image dehazing requirements.Aimed at the color cast produced in the process of imaging or haze removal,the color cast detection method based on Lab space and the correction method based on histogram equalization is proposed.Finally,we use the c ++ language with VS2012 and Opencv,the video image dehazing algorithm is realized.Through a series of experimental results,this paper proves the effectiveness and real time of video image enhancement and dehazing method based on dark channel prior.