Research On HDR Panoramic Image Synthesis Technology

High dynamic range(HDR)panoramic image technology is a new image technology which combines HDR image technology with panoramic image technology to generate panoramic image with high dynamic range.HDR panoramic image can perfectly describe the actual content of the real scene,so this image technology has broad application prospects,such as virtual reality(VR),medical imaging,game development,live video and so on.HDR panoramic image technology has great value,but the existing HDR panoramic image imaging technology is still immature,which greatly limits the development of HDR panoramic image technology.Therefore,this dissertation has carried out research on HDR panoramic image synthesis technology,mainly including the following three original works:(1)Image stitching is a hot topic in computer graphics research,the representative As-Projective-As-Possible(APAP)stitching algorithm segments image into uniform blocks to achieve local image warping,which does not take full account of the differences between different image regions.In order to balance the accuracy and the efficiency of image stitching,an image warping algorithm based on the feature points distribution is proposed,which uses quadtree to segment image into blocks to achieve local image warping for image stitching.Firstly,the maximum depth of the quadtree is set,and the matching threshold of each image block is estimated according to the amount of image matching points.Then,we use the maximum depth and the matching point threshold to adaptively segment the image into blocks by quadtree.Finally,homography matrices of different image blocks are estimated by Moving Direct Linear Transformation algorithm,according to which image stitching is completed.The experimental results show that the proposed algorithm increases the accuracy of image stitching,and the speed is much faster than the APAP algorithm,and the quality of the stitched image is improved.(2)Multi-view multi-exposure image acquisition method captures a lot of redundant information,which reduces the efficiency of subsequent HDR panoramic image synthesis.In order to solve this problem,a single camera dense viewpoints multi-exposure acquisition is proposed.The feature of this method is that only one image is collected at each viewpoint,and the overlap of viewpoints between adjacent viewpoints is higher than that of multi-viewpoint multi-exposure method,and the exposure of images collected by adjacent viewpoints is different.The experimental results show that the method is not only simple and feasible,but also effectively reduces the redundant information generated in image acquisition,greatly reduces the workload of subsequent HDR panoramic image synthesis,and doubles the efficiency of HDR panoramic image synthesis.(3)In HDR panoramic image synthesis technology,multiple image registration and fusion are generally required,which can easily cause image distortion.In addition,the general HDR panoramic image synthesis method can not deal with the image sequence acquired by single camera dense viewpoints multi-exposure acquisition technology.To solve these two problems,an HDR panoramic image synthesis algorithm based on common reference image is proposed.Firstly,a common reference image is selected;then,all the images are registered to the common reference image;then,the camera response function is calculated using multi-exposure image;finally,all the images are transformed into the irradiation domain using the inverse camera response function,and the irradiation domain image fusion is carried out to obtain a HDR panoramic image.The experimental results show that the algorithm reduces the image distortion caused by image registration and image fusion,and can synthesize better HDR panoramic images.Moreover,the algorithm has wide applicability and can process image sequences acquired by three image acquisition techniques simultaneously.