Research on image dehazing method based on the generalized Gaussian function

In the rapid development era of digital information, images, as an important carrier of information transmission, are widely applied in various fields. Thus it is important to find ways improving the acquired image quality under adverse weather conditions such as haze. When light propagates in a foggy environment, it is scattered and absorbed when encountering tiny fog-particles. And it will result in a decrement of brightness and loss of details for the acquired images. Moreover, when the visibility limit is less than 1km, the effect of scattering loss will be dominant. Based on this, this paper proposes a de-attenuation method based on the generalized Gaussian function. This function has the ability to flexibly adjust its shape and can accurately fit the attenuation characteristics of different foggy scenes. Through de-attenuation experiments on a large number of foggy image datasets and using the method of comparative analysis, the de-attenuation effects of the generalized Gaussian function under different parameter settings are studied. In this process, a mapping relationship between foggy scenes with different visibility levels and their optimal de-attenuation generalized Gaussian functions is established. And thus optimal defog effects are achieved.