Research On Tone Mapping Algorithm Of High Dynamic Range Image Based On Feature Driven

With the rapid development of information technology,digital image processing technology has been widely used in machine vision,medical image processing,remote sensing image processing and other fields,and people's requirements for image quality are also increasing,while the description of real natural scenes in traditional low dynamic range images is very limited.At present,people usually use high dynamic range image recording light value of natural scene,most of the display device can only display LDR images,then how to solve the problem of dynamic range of the HDR image display device does not match became the focus of attention.Therefore,researchers have proposed a tone mapping algorithm to compress the dynamic range of HDR images to the applicable range of display devices.The main work of this paper is as follows:(1)The basic theoretical knowledge of high dynamic range images is briefly introduced,and four tonal mapping algorithms based on gradient field compression,bilateral filtering,weighted least square filtering and local edge-preserving edge filtering are analyzed and compared.(2)After theoretical research on the multi-scale tone mapping method,using the characteristics of guided filtering and Hessian matrix,a tone mapping algorithm combining guided filtering and nonlinear second-order features is proposed.The algorithm first obtains the brightness information of the input image,the brightness image is decomposed into base and detail layers of different scales.A nonlinear second-order feature method of the Hessian matrix is used to extract the high-frequency information of the base layer,and then the initial weight mapping of the base layer and the detail layer is constructed,and then carried out by guided filtering.After correction,the adaptive weight coefficients of the base layer and the detail layer are calculated,and the brightness image is recombined,and finally the color processing is performed.The hierarchical design of the algorithm is more reasonable,and the problem of setting parameters based on experience is avoided when the brightness image is reconstructed.Experimental results show that the algorithm can improve the local contrast of the image and the details of the light and dark areas are highlighted.(3)In order to design a better filtering algorithm,a total variation(TV)model is introduced.A high-order edge-preserving total bending(TB)model is designed on the basis of the TV model,and a tone mapping algorithm based on the high-order edge-preserving model is proposed.The idea of the algorithm is to use the TB model to decompose the brightness image to obtain a segmented smooth base layer image.The detail layer is obtained by subtracting the brightness image from the base layer image.The base layer contains the low-frequency information of the image.After compressing its dynamic range,the details are combined.The layers are combined to obtain a reconstructed brightness image,and the brightness image is color restored to obtain the final result image.The experimental comparative analysis shows that the base layer image decomposed by TB model can well retain the image edge,effectively remove the image noise,make the color bright,and restore the image details and texture.