Image Enhancement For Hazy Scenes Using Modified Dark Channel Prior

The age of image enhancement is as old as the image itself.Noise is annoying or unwanted data in image information,which is required to be removed.The types of noise in images vary with technology,image capturing equipment,image transmission media,and environmental effects.A lot of image enhancement techniques exist in literature as there is no general enhancement method that may apply to remove all types of noise.One of the problems faced by photographers during capturing of images is the hazy environment.It deteriorates visibility,color,and crispiness of a scene.Moreover,it also degrades the performance of many potential image processing applications,which may be working in critical systems such as uncrewed aerial vehicles,surveillance and control,visual object tracking,etc.This thesis is aimed to propose a new method for enhancement of hazy images.The work introduces the degradation in images due to cloudy,foggy,or smoggy weather and discusses the enhancement methodology of such images.Different methods for evaluation of haze removal algorithms are explained in this work,and various methods for enhancement of images affected by haze are also elaborated.Finally,an image enhancement algorithm is proposed for pictures taken in hazy,fogy,smoggy environment.The algorithm is based on model base enhancement strategy in which noise model is considered first and then from the model,the enhancement method becomes easy as we inverse the model to find out the clear picture.The proposed method is tested on various images,and results are also compared with existing image enhancement methods.Simulation results showed that the recommended method is better for contemporary approaches both in terms of qualitative and quantitative measures.The proposed method is also computationally cost-effective,which is the desirable feature for online image enhancing algorithms.