Enhancement Of Degraded Image Based On Atmospheric Scattering Model

Images acquired in bad weather,such as haze,fog and sandstorm are seriously degraded by the scatting of the atmosphere,which causes image color distortion,reduces the contrast and makes the object features difficult to identify.Bad weather not only reduces the visual effect of the image,but also makes it difficult to the image procession.The bad weather also leads to the inconvenience of all kinds of instruments which rely on optical imaging,such as satel ite remote sensing system,aerial photo system,outdoor monitoring system and object identification system.Therefore,it is important to enhance the degraded image and improve its visual effects and quality.In this paper,enhancement of degraded image based on atmospheric scattering model in haze and sandstorm weather is discussed.The major work can be summarized as follows:(1)According to Dark channel prior in RGB color space,the dark channel prior's representation in HSI color space is derived.In this paper,the performance of the dehazing algorithm based on dark channel prior is analyzed.Aiming at the problem of color distortion and the low efficiency in the dehazing algorithm base on dark channel prior,a dehazing algorithm based on HSI color space is proposed.The proposed algorithm estimates air light and rough transmittance in the HSI color space and use the guided filter to refine the rough transmittance by the threshold method.Then,restoring the image based on atmospheric scattering model and using tone mapping to get the haze free image.Compared with the experimental results of various defog algorithms,the algorithm proposed in this paper is more efficient and has better effect.(2)The influence of atmospheric particles on the quality of optical imaging and the reason of color shift of sandstorm image are analyzed.An algorithm of sandstorm image enhancement based on HSI color space is proposed.According to the hypothesis of gray world,the histogram specification is used to color the sandstorm image,and obtain the general haze image.Then,the obtained image can be enhanced by using the dehazing algorithm based on HSI color space.Experimental results show that the proposed algorithm can effectively eliminate the color shift phenomenon and improve the clarity and contrast of the sandstorm image.(3)A dehazing algorithm based on machine learning is proposed.In this paper,the haze image-transmittance map is constructed by using the image dehazing algorithm based on HSI color space.The training method is used to train the image library to obtain the transmittance's visual word which get the transmittance visual dictionary.The visual dictionary is used to detect the haze images and then get their corresponding transmittance maps.Quartet is used to estimate the atmospheric light and the images are restored by the atmospheric scattering model.Experimental results show that the machine learning and training method can be used to dehaze,and the restored image has good results.