The Algorithm Of Multi-exposure Image Fusion Based On Detail Enhancement And Ghosting Removal

With the rapid development of digital imaging technology,the requirement of visual aesthetics are also increasing.The development of high dynamic range imaging technology has greatly promoted the development of digital images in the direction of high quality and high information.At the same time,this technology has been widely applied to various fields of life.As an effective method to synthesize high dynamic range image,multi-exposure image fusion technology has become a research hotspot at home and abroad.According to different shooting scenes,it can be divided into multi-exposure image fusion based on static scenes and multi-exposure image fusion based on dynamic scenes.Therefore,how to obtain the fused image with clear details on the basis of the minimum visual distortion,and how to eliminate ghosts caused by moving objects in dynamic scenes have become a difficult problem to be solved in the current multiexposure image fusion technology.The related theories and methods of multi-exposure image fusion were studied in this thesis,that keep preserving the details of bright and dark areas as much as possible and effectively eliminating ghosts as research focus.this thesis proposed an improved multi-exposure image fusion algorithm based on static scenes and dynamic scenes to improve the quality of fused images.The research content of this thesis mainly includes:1.Aiming at the problem that traditional multi-exposure image fusion algorithm keep the pixels participated in image fusion concentrating at about 0.5,which leads to the incomplete retention of details in the final fused image.the exposure brightness weight function was improved in this thesis and brighter pixels and darker pixels was preserved on the basis of eliminalting over-exposed and under-exposed pixels,so that more effective information in the original image participate in image fusion.2.Aiming at the problem that high dynamic range image's texture details are not clear synthesized by the multi-exposure fusion algorithm,a three-scale fusion framework was proposed by circular mean filtering,dividing image into base layer,middle layer,detail layer,then the weight map of the three-scale image was calculated combining the guided filtering,and the final fused image is obtained after weighted sum fusion.3.Aiming at the problem that ghosts are generated by directly merging images in dynamic scenes,ghost detection algorithm for double detection motion bitmaps was proposed,which is used for multi-exposure image fusion in dynamic scenes.Firstly,the median exposure image was selected as the reference image,and the initial motion bitmap of the image to be fused was calculated according to the pixel order relation.Then,the initial motion bitmap was corrected according to the brightness consistency,and the final motion bitmap was obtained.The motion bitmap was used as one of the weighting items of the image fusion,combining the other three weights,and the fusion weight map was calculated,the fused image without ghosting was obtained after weighted sum fusion based on three-scale image fusion framework..