Research On Dehazing Algorithm Based On Linear Transmission And Color Space Conversion

Nowadays,intelligent products have penetrated into all aspects of our lives.In this context,the application scenarios in the field of computer vision have also improved,such as target detection and recognition,unmanned driving,community monitoring,pedestrian and vehicle detection systems,intelligent transportation,etc.However,the application of this science and technology is easily affected by weather conditions,especially in severe weather such as fog and haze.Under fog and haze weather conditions,tiny particles suspended in the air and large-diameter aerosol particles will affect the light Due to different degrees of reflection,refraction and absorption,the images captured by imaging devices often have problems such as color distortion,decreased contrast,blurred images,and dimness.It seriously affects the subsequent intelligent processing of images.Therefore,image sharpening processing of foggy scenes plays a crucial role in computer vision applications.This paper studies three types of algorithms for image sharpening processing under foggy conditions at present,and focuses on the single image restoration algorithm based on physical model,and deeply analyzes algorithms such as dark channel prior,color attenuation prior and linear transmission.Aiming at the existing problems,two image dehazing algorithms are proposed.(1)Aiming at the problem of blocky effect and inaccurate estimation of transmittance in the sky area in the dark channel prior algorithm,an iterative optimization dehazing algorithm based on improved linear transformation is proposed.Firstly,starting from the minimum color component,an adaptive linear transformation model of the atmospheric dissipation function and the minimum color component is established to obtain the initial estimate of the medium transmission rate;then,the medium transmission rate in the bright area of the fog map is adaptively corrected;Secondly,the high-order filter is used for iterative optimization to obtain the optimal medium transmission rate;finally,the optimized medium transmission rate and the atmospheric light value obtained based on the quadtree sub-block search method are substituted into the atmospheric scattering model to obtain the restored image.The experimental results show that the algorithm can restore the image clearly and naturally,and has a better dehazing effect on a single image.(2)In order to solve the problem of inaccurate transmittance estimation in the single image dehazing algorithm and failure to select the atmospheric light value when there are white bright objects such as the sky in the restored image,a non-linear transformation dehazing algorithm based on YCr Cb color space is proposed.First,the nonlinear transformation model of the Y component of the hazy image and the Y component of the non-haze image in the YCr Cb color space is obtained from the experiment and theoretical analysis;then,the relationship between the components in the RGB and HSV color spaces is used to construct the The adaptive weight coefficient of the model;secondly,the threshold method is used to obtain the atmospheric light value in the YCr Cb color space;finally,the brightness of the restored image is adjusted according to the Weber-Fechner law.The experimental results show that this method has certain advantages compared with the existing dehazing methods.