Research And Application Of Image Dehazing Algorithm

In recent years,due to the influence of haze weather,the video images collected by image acquisition equipment have been seriously degraded,which has serious impact on applications such as intelligent transportation,aerial surveying,driving assistance and video surveillance.The image dehazing algorithm can recover clear fog-free images from foggy images,effectively improving the sharpness of the image.It is of great significance to optimize the processing of video images captured by the security monitoring vehicle system under the influence of illumination and extreme weather,and to reconstruct more identifiable monitoring materials and as the pre-processing of image detection and recognition in the field of computer vision.Based on Retinex theory and atmospheric scattering model this paper proposes a smog image enhancement algorithm based on fractional differential and multi-scale Retinex joint and an image dehazing algorithm based on gray correlation guided filtering.The smog image enhancement algorithm based on fractional differential and multi-scale Retinex joint firstly processes the original image with the fractional differential algorithm and transforms the image from the RGB color space to the HSI color space.The Gaussian filter in the multi-scale Retinex algorithm is replaced by a pilot filter,and the saturation layer is enhanced by the gamma correction function.Finally,the processed image is converted into an RGB image,thereby obtaining an enhanced restored image.The image dehazing algorithm based on gray correlation guidance filtering is inaccurate for the fog defocusing algorithm based on the image dehazing algorithm based on atmospheric scattering model,which leads to the problem of dark color and halo artifact after image defogging.Firstly,the gray correlation theory is applied to judge the pixel value of the haze image.The pixels of the foggy image are divided into normal pixels and pixels destroyed by haze particles.Then,the pixels damaged by the haze particles are guided and filtered.The atmospheric veil value is estimated in the number domain;finally,a clear image is recovered based on the inversion of the atmospheric light scattering model.The experimental results show that the two algorithms proposed in this paper can not only effectively improve the sharpness of the haze image,but also solve the problems of darkness,color distortion and halo in the sky after defogging.The image defogging algorithm based on gray correlation guidance filtering is used to monitor the defogging of video images.The feasibility and good adaptability of the dehazing algorithm are verified.A system architecture using FPGA to implement the algorithm is designed.The algorithm is applied to practical engineering and provides a high-level engineering application value for traffic monitoring intelligent construction.