Professor Sogang University Seoul, Republic of Korea
High dynamic range (HDR) techniques have received significant attention in generating realistic, high-quality images and videos and improving visual quality in new display systems. We have witnessed remarkable advances in HDR reconstruction using deep learning technologies in recent years. This review examines recent developments in HDR reconstruction using a deep learning approach, which takes a single low dynamic range (LDR) image as an input and aims to restore an HDR image featuring higher color gamut and a higher detail retention than the LDR image. We aim to provide a comprehensive survey in this field. Since there are numerous HDR algorithms, it is necessary to evaluate and organize their performance; therefore, we evaluate them using two objective evaluation metrics.