Research On Image Defogging Algorithm Based On Dark Channel Prior

Nowadays, the haze weather will often appear in most of city. Fog is not only hazardous to human health, but also causes a great impact on people's lives,equipment use and so on. Such as, fog and haze weather leads to traffic jams and obvious decline of the quality of outdoor monitor images. In fog and haze, most of video monitor image's color and contrast will become dim. So this phenomenon leads to some important details submerged in fog difficult to find, causing the normal function of the video surveillance system will be affected. Intelligent transport system is based on video surveillance system, which undoubtedly affects the development of intelligent transport system. So there is a great important significance in research of fog image sharpening.Firstly, based on intelligent transportation system background of the topic, the paper discusses the development of intelligent transportation system and image defogging technology, analyzes the challenges of image defogging technology.Secondly, the paper introduces the classification of image defogging algorithm,which is image defogging algorithm based on image enhancement and physical model.It elaborates the basic principle of contrast enhancement algorithm, histogram equalization algorithm, Retinex algorithm, Tan algorithm, Fattal algorithm and He algorithm, and gives some related experiments. Through experiment simulations, the paper analyzes the advantages and disadvantages of these algorithms and application scenario, and discusses image quality evaluation index for these images after defogging, that is, from the perspective of subjective and objective.Then, in the view of the image dark channel prior, the paper detailly analyses the principle of atmospheric scattering model, and states the specific processes of dark channel prior algorithm, whose core part is to optimize the transmission. Because the complexity of soft matting and guide filter is high and its running time is too long, the paper proposes an image defogging algorithm based on dark channel prior and morphology. Firstly, this paper estimates the global atmospheric light's value and rough transmittance. Secondly, it optimizes the rough transmittance by morphological open operation. Then we take some calculated values into simplified model to acquire the defogging image. Finally, given that the image brightness reduces after using the dark channel algorithm, this paper eventually improves the image brightness. Theexperimental results show that this algorithm can effectively and quickly recover the degrade image under fog weather, which has a lot of improvement in processing speed.Finally, the paper completes the construction of software platform of image defogging system, and designs the image display and saving control, image information, opening file, image defogging and other function models, then introduces the development and implementation process of functional modules of software system in detail.