Research On High Dynamic Range Image Fusion And Display Technology

Due to the constraints of the current hardware level,the image taken by ordinary cameras usually suffer from under-exposed or over-exposed.It can't record the real information of the scene observed by the human eye,because of its restriction in brightness,contrast,color and other aspects.High dynamic range images can cover all the dynamic range of the scene,which has a better effect in terms of color saturation and detail information,and it could bring the viewer a more realistic sensory experience.High dynamic range technology has enjoyed wide application in recent years,such as satellite remote sensing,medical imaging,video surveillance,3D game industry,and achieved good application effect.This thesis mainly focuses on the relevant technologies of high dynamic range.First,it introduced and analyzed the related technologies used in imaging and display,including high dynamic range imaging technology,image registration,ghost removal,multi-exposure image fusion and display technology of high dynamic range image.This paper discussed the development status of each aspect and analyzed the advantages and disadvantages of existing technology.Then,this paper focuses on the multi-exposure images fusion technology.According to difference of fusion methods,it can be divided into two kinds of high dynamic range image generation method.One is based on pyramid transformation and another way is to directly fuse multi-exposure images.The fusion method based on pyramid fuses the multi-exposure images at different scales and different resolutions,so we can get the high dynamic range images with rich details.This paper mainly studies the two kinds of fusion algorithms which are based on Laplacian pyramid and gradient pyramid,and then analyzes the advantages and disadvantages of each method through experiment comparison.The direct fusion method usually uses different method to generate a weight map,and then carries on the weighted fusion to the multi-exposure image sequences.In this paper,we proposed an effective weight map generation method,which was done pixel by pixel under the lead of 2 parameters: the amplitude of gradient and luminance level,and then analyzed the effect of this method with multiple sets of image sequences.Experimental results show that the proposed algorithm can effectively detect the best exposure pixels in the image sequence,especially for the dark region of the image.In this paper,we evaluatethe experimental results in the subjective and objective aspects,and the evaluation results show that the algorithm can effectively expand the dynamic range of the image.The fusion result has achieved good effect in color saturation and pixel resolution.