Research On Tone Mapping Algorithm For High Dynamic Range Images

High dynamic range images are stored in high-order data such as 16-bit and 32-bit,which preserves the real scene in nature and has a wealth of detailed information.However,the common display device is 8-bit,so high dynamic range images need to be tone mapped to be displayed on a normal display.The task of this paper is to realize the tone mapping of high dynamic range images,so that the compressed image can maintain more detailed information as much as possible,to keep the corresponding color information and avoid visual artifacts such as halo phenomenon,which is convenient for later human eye observation or machine processing.The main work and research results of this paper are as follows:1.The commonly used edge-preserving filters are studied,analyzed and compared:bilateral filtering,weighted least squares filtering,₀L smoothing filtering and local edge-preserving filtering.The tone mapping algorithms based on edge-preserving smoothing filter is studied.Then the reason of the halo phenomenon is analyzed.Finally,the weighted least squares filter is selected through experiments and comparative analysis.2.Aiming at the problem that the details of the tone mapping algorithm improved by Khan et al which is based on histogram and human visual model are not enough,a detail-maintained tone mapping algorithm TWT is proposed.The algorithm first converts the image into a brightness channel.The decomposition of the image base layer and the detail layer is performed by weighted least squares filtering,then compress the base layer using a histogram and human visual model to create a lookup table and enhances the detail layer by S curve.After the merge process,the process of removing noise and color correction is performed.By doing this to achieve high bit width tone mapping and keep the details.The subjective and objective evaluation of the contrast experiment shows that the algorithm maintains the image detail information better than the Khan algorithm,especially the highlight and low-dark information.At the same time,the algorithm can maintain the color information of the image without gradient inversion and halo phenomenon,and the tone mapped images quality are higher.3.Aiming at the problem that the time keeping complexity of the algorithm is high,a tone mapping algorithm TGT based on gradient domain guided filtering is proposed.Based on the guided filtering,the algorithm adds explicit first-order edge perceptual constraints to better preserve the image edges and reduce the halo phenomenon.By doing this to realize the decomposition of the basic and detail layers of the image during the high bit width mapping.The running efficiency of the algorithm is at least 50%higher than that of the TWT algorithm.The subjective and objective evaluation of the experimental image shows that the image quality obtained by the algorithm is basically the same as that obtained by TWT algorithm.