High Dynamic Range and Super-Resolution Imaging From a Single Image

• Main Authors: Jae Sung Park, Jae Woong Soh, Nam Ik Cho
• Format: Article
• Language: English
• Published: IEEE 2018-01-01
• Series: IEEE Access
• Subjects: Image enhancement; high dynamic range imaging; super resolution; convolutional neural network
• Online Access: https://ieeexplore.ieee.org/document/8267234/

This paper presents an algorithm for high dynamic range (HDR) and super-resolution (SR) imaging from a single image. First, we propose a new single image HDR imaging (HDRI) method based on the Retinex approach and exploit a recent single image SR method based on a convolutional neural network (CNN). Among many possible configurations of HDR and SR, we find an optimal system configuration and color manipulation strategy from the extensive experiments. Specifically, the best results are obtained when we first process the luminance component (Y) of input with our single image HDRI algorithm and then feed the enhanced HDR luminance to the CNN-based SR architecture that is trained by only luminance component. The ranges of chromatic components (U and V) are just scaled in proportion to the enhanced HDR luminance, and then they are bicubic interpolated or fed to the above CNN-based SR. Subjective and objective assessments for various experiments are presented to validate the effectiveness of the proposed HDR/SR imaging scheme.