부산대학교 도서관

The image signal processing (ISP) pipeline reconstructs compelling RGB images from raw images. Recently, deep neural networks have been proposed to replace conventional divide-and-conquer ISP pipeline solutions. But previous methods cannot control the luminance level necessary for the camera system and suffer from severe color distortion in output images. In this paper, we propose a controllable luminance neural camera based on a convolutional neural network (CNN). Experiments show that our method can achieve flexibility in adjusting luminance levels and higher fidelity in restoring color.