검색 상세

마스크 적용 신경회로망 기반 톤 맵 하이 다이나믹 레인지 영상 복원 기법 연구

Tone-mapped High Dynamic Range Image Restoration with Mask-applied Neural Network

초록/요약

This thesis proposes a neural network-based image restoration method for recon-structing a single tone-mapped high dynamic range image, which has a higher color reproduction rate and higher detail preservation than a low dynamic range image, from a single low dynamic range image including over/under exposed images. Tone-mapped high dynamic range image restoration problem can be regarded as a combination of existing ill-posed problems: (1) image color restoration and (2) image detail restoration. To solve these two ill-posed problems at the same time, this thesis uses a mask and a mask-applied neural network to distinguish between clipped and non-clipped regions. And this thesis restores the clipped region based on an image inpainting viewpoint and the non-clipped region based on image-to-image translation viewpoint. The proposed method showed lower color restoration ability for the non-clipped region, but showed higher color and detail restoration for the clipped region within a certain size than conventional methods, despite the relatively small number of neural network parameters. In addition, the qualitative results showed that the proposed method restores the clipped region naturally with-out degrading the perceptual quality compared to other methods.

more