Research on single image haze removal algorithm based on parameter optimization search of linear model

Abstract

Image defogging method based on non physical model (the image contrast enhancement method) and image dehazing method based on physical model (the atmospheric scattering model method), it is the two kinds of method that is focused on at present. The innovation of this article is: the estimation value of parameter achieved by atmospheric scattering model image dehazing method is not optimal; and the method of contrast enhancement makes the part of clear image and the depth of information distortion. On the basis of the theoretical analysis this paper gives an improved linear model which is proposed to describe the fog removal principle based on the atmospheric scattering model and the contrast enhancement method, and the fog removal problems of models were converted into the optimization searching in parameter space. Initial estimates parameters of atmospheric scattering model method based on linear model in the aspect of solving the problem, and the genetic algorithm and the clarity decision function are introduced to optimize the parameters. And then obtain defogging image. Experimental results demonstrate that the proposed method can achieve a better dehazing effect and applicability compared to some state-of-the-art methods.