The Research And Implementation Of Image Processing Methods In Fog And Haze Environment

Haze,which is the result of interaction of human activities under specific climatic conditions.In this society,human social activities and economic development are bound to discharge a large number of fine particles.If the discharge amount exceeds the circulation and carrying capacity of the atmosphere,the concentration of fine particles will continue to accumulate,meanwhile haze will appear under severe weather.The contrast of images will be reduced by haze.The technique of removing haze from images plays an important role in various fields,such as traffic navigation,object recognition,military affairs,urban road transportation,intelligent transportation system,remote sensing system.In terms of haze removal algorithm of a single image,this thesis mainly focuses on the following work:1.After studying and applying DCP(Dark Channel Prior),the haze removal algorithm of a single image,it is found that there is block effect in the original haze transmission,leading to halo in the recovery results.In order to increase the speed of optimization,we have improved the optimization algorithm.2.This paper has designed a quicker single image haze removal algorithm based on the DCP prior knowledge and atmospheric scattering light.First the global atmospheric light is calculated using the DCP prior knowledge.Second,the airlight is estimated.Third,the global atmospheric light and airlight are used to calculate the haze transmission rate.This transmission rate does not need to be optimized.Besides,the recovered images can be directly obtained using the atmospheric attenuation model.Our algorithm combines both the advantage of the accurate estimation of global atmospheric light by DCP and the quick calculation of airlight,which makes it 10 times faster,meanwhile the recovered images are similar to or even better than that using DCP.