Research On Enhancement Algorithm Of Specular Highlight Image Based On Dark Channel Prior

In real life,the image affected by specular noise is very common.In order to improve the quality of the image and reduce the interference of specular noise,researchers have proposed many methods to remove the specular component in the image.However,the traditional specular processing methods not only need to set a lot of priori conditions,but also need a lot of image assistance,and the image quality after processing by these methods has the problem of decline.Therefore,according to the different input data of previous experiments,this paper proposes an enhancement technology for single specular image.This method mainly deals with high-light images in real life scenes,which greatly improves the application scenarios and scope of such methods.The main research contents and contributions of this paper are as follows:Firstly,a new method to enhance the texture information of specular image is proposed by studying the specular image in real life.Firstly,in the theoretical part of this paper,we analyze the scene features of the highlight image.Then two kinds of specular images with different scene features are obtained,namely,the specular images in the laboratory scene and the real scene.Finally,specular high-light images taken in real scenes are determined as the research object of this paper.Secondly,this paper presents a fast and effective method to enhance specular image.Firstly,the image is converted from RGB space to HSV space by using the thinking of spatial transformation.Then,V component in HSV space is adjusted to adjust H and S components in an adaptive way.Then the guide filter algorithm is used to process the image,and the highlight image is divided into the basic layer and the detail layer,and the image information of the two layers is processed respectively.The main contents are as follows: the adjustment parameters based on local variance are introduced into the cost function of the base layer to better maintain the edge information of the image in the process of processing;Then in the detail layer,the amplification factor of the detail layer is solved in an adaptive way,in order to enhance the image details and balance the noise better.Finally,the above two parts are superimposed to output the enhanced image.This method can effectively enhance the texture information in specular image.Thirdly,the dark channel priori algorithm is proposed to enhance specular image for the first time.In this method,an improved guided filter algorithm is used to optimize the rough estimation of the transmittance in the process of finding the dark channel of the image,and then a new method is proposed to estimate the atmospheric light value.Finally,the local pixel based weighting function is introduced in the boundary constraint,and the CLAHE algorithm is used to adjust the L-component,A-component and B-component of the image in LAB space.Compared with the existing specular removal algorithms,this method does not require any prior assumptions.Finally,a specular image processing and analysis system is developed based on this algorithm.