Research On The Image Clear Processing Under Sand-Dust Weather

Under the sand-dust weather,the suspended sand particles in the air have an effect on the absorption and scattering of the light,resulting in degradation of the collected images,which are characterized by color distortion and decreased clarity.This has seriously affected the normal operation of intelligent systems which rely on computer vision,such as city security monitoring and road traffic monitoring.In order to solve these problems,this paper studies the clearing method of the image obtained in the sand-dust weather:1.Aiming at the problem of color deviation and low contrast of sand-dust weather image,an image sharpening algorithm is proposed from the perspective of image restoration.The algorithm consists of two parts: the color correction based on Gaussian model and the contrast enhancement based on bright and dark channels.Firstly,according to the distribution characteristics of RGB channels of histogram of the sand-dust image,the Gaussian model is used to model each channel separately,so as to adjust the image color.Secondly,the pixel-level atmospheric light value is calculated based on the dark channel and the bright channel,and then the transmission image and atmospheric light image are optimized to restore a clear image without sand and dust.Finally,the experimental results show that the proposed algorithm can effectively improve the visual quality of sand-dust image,but it is not effective in processing sand image with severe information degradation.2.Aiming at the problem of information loss of blue channel,low contrast and uneven brightness in severely degraded dust weather images,a sand image sharpening method based on Lab color space is proposed from the perspective of complementary advantages of image restoration and image enhancement,which can effectively enhance sand-dust images of different degrees.Firstly,the a-b channel in the Lab space is used for color correction of sand-dust image,and then stretch the color in RGB space,so as to effectively solve the problem of sand-dust image color deviation of various degrees.Secondly,the dark channel principle is used to enhance the contrast of sand-dust image.Then,in the Lab space,the color channel of the contrast enhanced image and the brightness channel of the original sand-dust image enhanced by Retinex theory were recombined to obtain a clear image.Finally,the effectiveness of the proposed algorithm is verified from two aspects of subjective quality and objective evaluation by experimental comparison with the existing sand-dust image clearing methods.