Research Of Algorithm Of Haze Image Sharpening

With the continuous development of image processing technology,computer monitoring system and imaging system can capture a variety of image information.However,haze and other bad weather conditions will bring many challenges to the image acquisition system,causing the captured images to degrade and reduce their value.Therefore,it is of great theoretical significance and application value to research on the pictures acquired by the image acquisition system under haze and other weather conditions.Based on this background,this thesis studies the existing defog algorithm theory and key technologies,then proposes a Retinex defog algorithm based on dark primary color and homomorphic filtering.The main work and achievements are as follows:Firstly,the current research status of defog algorithm is summarized and classified.The research results of Retinex algorithm in the defog field are mainly discussed.Secondly,studying the characteristics of foggy images,analyzing and simulating the existing classic fogging algorithms,summarizing the existing problems and giving the evaluation standards of image quality.Thirdly,this paper proposes a Retinex algorithm based on the dark channel priori theory.First optimize the dark channel priori algorithm to obtain the initial defogger image,and the image is spatially converted to establish the exponential relationship between the Gaussian scale parameters and the optimized transmissivity,reconstruct the Gaussian filter core to improve the MSR algorithm and achieve large-scale transformation of near scene to enhance color fidelity,make small-scale transformation of distant scene to enhance the image detail display in order to achieve the adjustment of the brightness component I.The saturation component S is adaptively linearly stretched to optimize the color of the image,and finally the space is restored to obtain the final defogged image.The experimental results show that the proposed algorithm enhances the image sharpness and contrast significantly while achieving a good image color gradation.Fourthly,a Retinex image enhancement algorithm based on improved homomorphic filtering is proposed in this paper.Firstly,the noise in the haze image is removed by image denoising,and then the improvement of homomorphic filtering is achieved by constructing a new Butterworth filter to correct the image brightness disparity caused by the uneven illumination in the image.So as to make the original part of the image with a strong contrast between the light and dark areas smooth.The processed image is enhanced by the MSR algorithm and adaptively linearly stretched to improve the contrast of the image.Finally,based on CLAHE algorithm to complete the contrast adjustment of the image,to improve some of the regional details of the increase is not obvious,the overall perception of gray shortcomings Experimental results show that this algorithm can effectively enhance the image enhancement effect.Finally,the thesis summarizes the research work,analyzes the advantages and disadvantages of the proposed algorithm,and makes a prospect of the further research of the algorithm.