Adaptive Exposure Fusion for HDR Imaging

HDR images are usually generated by fusing sequence of images captured with variable exposure time. Exposure Fusion is a technique that directly fuses the exposure shots into displayable image. This thesis proposes a novel Exposure Fusion algorithm that directly fuses exposure bracketed shots into a displayable image. Most techniques targeted for direct fusion do not have an effective exposure control mechanism and are only designed for exposure sequence containing adequate number of exposure shots. The proposed algorithm offers a novel approach that adaptively adjusts its parameter for the best viewing experience even for exposure sequence that do not contain adequate number exposure shots. An online user survey showed that the proposed method gave consistent superior results when compare with other methods of similar nature. The survey results were also compared against state of the art evaluation metrics (TMQI and HDR-VDP) and the survey results contradicted the evaluation metric.