A Research Of Ghost-free High Dynamic Range Image Generation In Dynamic Scenes

With the development of digital imaging technology and the increasing demand for high quality digital images,the emergence of high dynamic range imaging technology has greatly improved the quality of digital images.This technology expands the dynamic range of the image,so that the obtained image can simultaneously contain the detailed information of the bright and dark regions of the scenes,and can more truly reflect the information contained in the actual scenes.A common method at present is to capture low dynamic range images of different exposures with multiple frames.These images have dynamic ranges in different regions of a real scene,and multi-frame images are synthesized by a software algorithm into a high dynamic range image.However,this method is only applicable to multi-frame images shot in a static scene.Most of the real scenes are dynamic scenes,and there are moving people and vehicles in the multi-frame images captured,which makes the method generate artifacts in a high dynamic range image.The ghosts greatly affect the quality of the image.Therefore,researchers began to focus on ghost-free high dynamic range image acquisition methods in dynamic scenes,and the key technique for generating ghost-free high dynamic range image is to detect ghost pixels.Based on the principle of multi-exposure high dynamic range imaging and ghost detection and removal,this paper presents two ghost-free high dynamic range imaging algorithms.One of the proposed methods is a ghost-free high dynamic range imaging method based on the adjacent frame difference method.The ghost region is found by comparing the relationship between the gray values of two adjacent frames,and then the adjacent ghost regions are separated to obtain the ghost region of each input image.By selecting an image as a reference image,the ghost regions between the reference image and the rest of the input images can be acquired,there is a problem that the high dynamic range image generated in the single-frame reference image ghost removal algorithm has poor detail information in the ghost region,and an image restoration algorithm is used to improve the details of the image after the moving object.A high dynamic range image is finally obtained by using a multi-exposure composite imaging algorithm that improves the composition weights.The second proposed method is a ghost-free high dynamic range imaging method based on gradient structure consistency.After selecting a frame of image as a reference image,then calculating the gradient structure similarity between the reference image and the rest of the input images,the area with ghosts is identified by an appropriate threshold,and then a multi-exposure fusion algorithm with improved fusion weights is used to generate ghost-free high dynamic range image.In fusion,a laplacian pyramid algorithm is combined to avoid stitching problem of the resulting image after fusion.The experimental results show that two kinds of ghost-free high dynamic range image generation methods are effective in removing ghost effects,and the obtained images have better quality.