Exposure fusion (EF) directly generates a low dynamic range (LDR) image with similar image quality to the high dynamic range (HDR) image by combining images of different exposures to overcome the limited dynamic range of common digital cameras. EF for dynamic scenes is challenging not only due to the difficulty in removing various ghost artifacts but also because of the requirement of preserving details, especially in the saturated regions. Recently, low-rank (LR) matrix completion (LRMC) has been shown to be efficient in separating the latent background from the sparse motion in the irradiance domain. However, the performance of LRMC models strongly relies on the estimated irradiance images. To address this problem, this paper proposes a novel EF method combining Retinex theory and the LRMC model. The proposed method consists of four steps. First, according to Retinex theory, an image is decomposed into illumination and reflection components. Second, the LRMC model is applied to the reflection component to generate the background reflection component and the sparse error. Third, the motion map is modeled as a Markov random field (MRF), integrating the sparse error with the ordering constraint across all the illumination components. Finally, all the illumination components are fused via a pyramid-based method, where the weight maps are defined based on the obtained motion map and illumination. The experimental results show that the proposed method outperforms the state-of-the-art methods particularly in preserving details in saturated regions.

High dynamic range (HDR) imaging, aiming to increase the dynamic range of an image by merging multiexposure images, has attracted much attention. Ghosts are often observed in a resultant image, due to camera motion and object motion in the scene. Low-rank matrix completion (LRMC) provides an effective tool to remove ghosts. However, user specification of the included or excluded regions is required. In this paper, we propose a novel HDR imaging method based on bidirectional structural similarities and weighted low-rank matrix completion. In our method, we first propose the bidirectional structural similarities containing forward-projection structural similarity (FPSS) and backward-projection structural similarity (BPSS) to divide each image into four groups: motion region, saturated region in the source image, saturated region in the reference image, and static and unsaturated regions. Then, the weight maps and the motion maps constructed based on FPSS and BPSS are introduced in the weighted LRMC model to reconstruct the background irradiance maps. Experiments are conducted on several challenging image sets with complex scene, and the results show that the proposed method outperforms three current state-of-the-art methods and Photoshop cs6 and is robust to the reference image.