부산대학교 도서관

Exposure fusion is a technique to fuse several differently exposed low dynamic range (LDR) images to an LDR image. The output image has more information than each of the input images, but it could suffer from loss of fine details. In this paper, we propose a detail-enhanced exposure fusion algorithm by introducing an L 0 norm based optimization in gradient domain. The proposed algorithm extracts fine details from a vector field that is generated by using the gradient fields of the input images and adds the fine details to an intermediate image that is fused by an existing exposure fusion algorithm. Experimental results prove that the proposed method can enhance fine details for fused images.