Research Of Generated Methods For High Dynamic Range Panorama Images

The dynamic range of real scenes in nature is very high.And the dynamic range of images captured by digital camera is much lower than the range that can be recognized by humans' eye.Thus,there may be quite bright and dark regions in the images captured in scenes with high dynamic range.And that would be more common for panoramas for the much wider view angle probably leads to the high dynamic range.In order to present panorama images with clear brightness level and rich details,the research of this paper focus on the generation methods of high dynamic range panorama images.With images captured at different view angles,a high dynamic range wide-angle panorama image can be generated after a series of image processing techniques.The generation methods of high dynamic range(HDR)image are studied first which mainly include two categories.One is based on the multi-exposure image sequence and the other is based on the one-exposure image.The paper mainly introduces exposure fusion method and BMD method which is proposed to avoid ghosting.For further study,a new method based on multi-level gamma-transformed images is presented,which can be used to transform LDR image to HDR image quickly.And the HDR image generated have higher image entropy and average gradient.Two approaches of HDR panorama generation are proposed.One is based on oneexposure images.First,one-exposure image at each view is transformed to HDR image by multi-level gamma-transformed fusion method.Then the transformed images are stitched to one HDR panorama.The second approach is based on interval-exposure images.The overlapped regions between images are fused into HDR image which is used to be a reference later.And then histogram specification is performed on original images at each view.The generated HDR effect images are with similar luminance level with reference image.Finally,all the generated HDR effect images are stitched together to an HDR panorama.