Study On The Method Of Clearing The Image Of Foggy

With the continuous improvement of modernization level, computer vision technology is more and more widely applied in people’s life. However, due to fog and other inclement weather, the image acquisition device is very susceptible, which always results in severe image degradation. Specific influence includes color distortion, degradation of contrast radio and definition. In particular, in recent years, fog and haze weather have been occurred frequently in China, therefore, in order to ensure the system to work properly, the study of defuzzification for fog-degraded image has great significance.Currently, there are two main categories of method to defuzzify fog-degraded images; enhancement method based on image processing and restoration method based on a physical model. Based on those two defogging method, this thesis studied the methods to clarify image degradation caused by fog and haze. Main reach and study includes:1) Firstly, we introduced atmospheric scattering model detailedly; Secondly, we elaborated the reason of image degradation under foggy weather conditions; At the end, we introduced evaluation criteria of subjective image quality and objective evaluation index.2) This paper analysed defogging algorithm for dark channel prior theory, and detailedly stated the selection of light from the atmosphere, estimation and optimization, and the restored image transmittance, and compared the defogging algorithm with other algorithms. Experiments proved that defogging algorithm for dark channel prior theory can clarify the picture, and edge information of scenery can be well preserved. Besides, combined with image quality evaluation, this thesis summarized the advantages and disadvantages of defogging algorithm for dark channel prior theory.3) Through the validation analysis for dark channel prior theoretical algorithm, this thesis improved algorithm based on image segmentation to fog. This algorithm improved aspects of atmospheric light selection, reflection optimization and real-time of the algorithm. The experimental results show that the algorithm is well performing for the scene of shallow fog. After restoring, the image detail and clarity is greatly improved, and the contours of the scene is highlighted, the time complexity is greatly reduced. The defogging idea aimed at image segmentation have much more purposiveness, and is more efficient.