Abstract:The reconstruction of high dynamic range (HDR) images from multi-exposure low dynamic range (LDR) images in dynamic scenes presents significant challenges, especially in preserving and restoring information in oversaturated regions and avoiding ghosting artifacts. However, current methods often struggle to address these challenges. To this end, our work aims to bridge this gap by developing a multi-exposure HDR image reconstruction (MHDR) network for dynamic scenes, complemented by single-frame HDR image reconstruction (SHDR). This network, comprising SHDR with enhanced stop image (SHDR-ESI) and SHDR-ESI-assisted MHDR (SHDR-A-MHDR), effectively leverages the ghost-free characteristic of SHDR and the detail-enhancing capability of ESI in oversaturated areas. Specifically, SHDR-ESI innovatively integrates SHDR with the utilization of ESI. This integration not only optimizes the SHDR process but also effectively guides the synthesis of multi-exposure HDR images in SHDR-A-MHDR. In this method, SHDR is specifically applied to reduce potential ghosting effects in multi-exposure HDR synthesis, while the use of ESI images assists in enhancing the detail information in the HDR synthesis process. Technically, SHDR-ESI incorporates a detail enhancement mechanism, which includes a self-representation module and a mutual-representation module, designed to aggregate crucial information from both reference image and ESI. To fully leverage the complementary information from non-reference images, a feature interaction fusion module is integrated within SHDR-A-MHDR. Additionally, a ghost suppression module, guided by the ghost-free results of SHDR-ESI, is employed to suppress the ghosting artifacts. Experimental results on four public datasets demonstrate the efficacy and superiority of the proposed method.

Divide and Conquer: Rethinking HDR Reconstruction Utilizing HSV Color Space

Dual-Focal Camera Hdr Imaging Based On Convolutional Neural Network

Single Exposure High Dynamic Range Image Reconstruction Based on Deep Dual-Branch Network

Ultra-High-Definition Image HDR Reconstruction via Collaborative Bilateral Learning

HDR Video Reconstruction with a Large Dynamic Dataset in Raw and sRGB Domains

High Dynamic Range Imaging with Context-aware Transformer

Deep SR-HDR: Joint Learning of Super-Resolution and High Dynamic Range Imaging for Dynamic Scenes

HDRNET: Single-Image-based HDR Reconstruction Using Channel Attention CNN

Self-Supervised High Dynamic Range Imaging with Multi-Exposure Images in Dynamic Scenes

Implementing evidence-based practice in Taiwanese nursing homes: attitudes and perceived barriers and facilitators.

HDR reconstruction from a single exposure LDR using texture and structure dual-stream generation

Learning Regularized Multi-Scale Feature Flow for High Dynamic Range Imaging

Single-Image HDR Reconstruction Assisted Ghost Suppression and Detail Preservation Network for Multi-Exposure HDR Imaging

HDR Video Reconstruction with Tri-Exposure Quad-Bayer Sensors

LHDR: HDR Reconstruction for Legacy Content using a Lightweight DNN

HDRTransDC: High Dynamic Range Image Reconstruction with Transformer Deformation Convolution

Efficient HDR Reconstruction from Real-World Raw Images

Progressive and Selective Fusion Network for High Dynamic Range Imaging