Algorithm And Research Of Severe Weather Image Enhancement

This paper analyzes the image structure and characteristics in daytime fog environment and night fog environment respectively.Through the establishment of mathematical model and cause analysis of normal light and low light images containing fog,the enhancement of fog images obtained by image acquisition equipment in different brightness environment is realized.The specific research contents are as follows:This paper studies the fog image defogging algorithm under normal light and the fog image defogging algorithm at night,expounds the relevant algorithms at home and abroad,analyzes the existing enhancement methods for fog images,and summarizes the image enhancement algorithms,mainly including histogram equalization,image gray processing,image smoothing,Retinex algorithm,image sharpening and wavelet transform.In order to verify the defogging and enhancement effect of the algorithm,the subjective evaluation criteria and objective evaluation criteria of the image are described respectively,so as to have a further understanding of the image quality.Aiming at the problem that the dark primary color a priori defogging algorithm does not estimate the transmittance of the sky region,resulting in block effect and color distortion in the restored image,a fog image enhancement algorithm based on sky segmentation is proposed according to the atmospheric scattering model.Firstly,the image is divided into sky region and non sky region through the improved threshold segmentation algorithm.The transmittance of the sky region is estimated and enhanced by combining the guided filter with the newly designed compensation function.Then,combined with the non sky region,the fog is removed by using the dark primary color a priori principle.Finally,in order to make the enhanced image more natural,the normalized weight algorithm is optimized to eliminate the edge residue at the decomposition.According to the classical incident light attenuation model and atmospheric light imaging model,the night fog image is abstracted as the combination of night image and fog image.For this image model,this paper proposes an adaptive brightness enhancement combined with improved Retinex defogging algorithm.The color space is used to decompose the image into brightness layer and structure layer,and the proposed adaptive brightness enhancement algorithm is used to optimize the brightness layer;Use the improved Retinex to process the structural layer.The experimental results show that the algorithm can restore the image with better brightness and contrast at night,and the image can be restored more effectively.In order to make the research algorithm more valuable,an image defogging enhancement system is developed.Through the analysis of image brightness,the adaptive selection algorithm is used to restore the image containing fog.Experiments show that the proposed fog clearing algorithm has good application value in practice.