Research On Adaptive Optimization Dehazing Algorithm Based On Linear Model

In hazy conditions,due to the scattering of atmospheric particles,images captured by image acquisition equipment suffer from contrast attenuation and color distortion,which severely affect people's visual perception and the performance of machine vision systems.In order to solve the degradation in haze conditions,it is necessary to study the theory and technology of dehazing.The thesis makes research on the foggy imaging model and the dark channel prior theory.Although the dark channel algorithm has achieved good dehazing effects,there are still many flaws,such as: the inaccuracy of the dark channel prior theory in the sky region leads to inaccurate transmission,the bright pixels and large sky region cause inaccurate estimation of atmospheric light,and the Halo effect in the restored image.The thesis focuses on the shortcomings of the dark channel algorithm and proposes two improved algorithms.Human vision perception and blind evaluation are used as the evaluation methods to analyze and evaluate the quality of restored image.The main research work of this thesis includes:(1)Aiming at the problem of inaccurate estimation of atmospheric light caused by highlight areas and large sky region,a dehazing algorithm combining dark channels and maximum channels of foggy images is proposed.Firstly,in order to solve the problem that the dark channel prior theory is not applicable in the sky region,a linear model is used to obtain the initial transmission.Then,the guide filter is used to refine the initial transmission to eliminate the block effect,blurred edges and details;Finally,atmospheric light is obtained by dark channel and maximum of the image.And then the image is restored by combining Atmospheric scattering model.Experimental results show that the image restored by the improved algorithm has appropriate brightness,natural color restoration,and more details of image.It achieves a good dehazing effect in different depth of fields.(2)Aiming at the problem of insufficient transmission and atmospheric light in traditional algorithm,an adaptive optimization dehazing algorithm based on linear model is proposed.Firstly,the edge information model is used to enhance the detailed information of the initial transmission map obtained by linear model to avoid Halo effects in restored images;Then,according to the dark channel prior,an adaptive optimized transmission is obtained to process the image including the depth of field region;Finally,the local atmospheric light estimation method is used instead of the quadtree method to avoid the inaccuracy of atmospheric light estimation in bright and sky region,and the image is restored by combining the physical model.Experimental results show that the improved restoration algorithm can recover rich detailed information.Due to the adaptive optimization of the transmission,the algorithm is also effective in processing images containing depth of field and dense fog images.