The Research And Application Of Single Image Defogging Algorithm Based On The Depth Of Field

Computer vision system has been widely applied in many fields, such asvideo monitoring, automatic driving, urban traffic system, etc. Under thecondition of fog and haze, the quality of images declines seriously. At thesame time, the images are so vague and dim and their contrast is poor. And theimages under the condition of fog and haze always appear especially seriousdistortion and color offset. Thus this situation limits the applications ofcomputer vision system.A highway fog warning system is designed and implemented through thesingle image defogging algorithm research. In this thesis, the main works aresummarized as the following four points.(1)Based on the research of the fog image degradation model and the darkcolor defogging algorithm, the deficiency of the dark color defoggingalgorithm is analyzed and several improvement methods are proposed.(2)Based on the analysis of the single image defogging algorithm, asingle image defogging algorithm based on the depth of field is developed.Firstly, a new physical model is established to eliminate the influence ofwavelength on image defogging and estimate atmospheric light intensity. Thenthe depth of field relationship ratio is used to evaluate image edge mutationarea. Moreover, bilateral filtering is utilized to achieve optimized transmissionchart. Finally, the fast defogging is carried out. Experimental analysis resultsshow that the algorithm has good robustness and calculates instantaneity.(3) In view of the image defogging evaluation uncompleted standard, theeffect of image defogging can’t be accurately evaluated. So, imagesdefogging evaluation system is developed, which evaluate the imagedefogging through the image edge information, color distortion, visibility,and some other aspects. The comparison shows that the evaluation system ofimages defogging is more objective and comprehensive.(4) On the base of above conclusions, the highway fog early warningsystem is designed with combining requirements of highway monitoringapplication. The characteristics of the video frames of video defogging canachieve highway fog visibility warning levels. The system run result provesthe effectiveness of single image defogging algorithm based on the depth of field.