BROWSE
Detail
Conference Paper
Adaptive Pre/Post Filtering on Self-Similarity based Super Resolution Video
Cited - time in 
Share 
Share
- Authors
- Dae Yeol Lee, Seyoon Jeong, Sukhee Cho, Jae-Hee Jun, Ji-Hoon Choi, Jong-Ok Kim, Hui Yong Kim
- Issue Date
- 2015-01
- Citation
- International Workshop on Advanced Image Technology (IWAIT) 2015, pp.1-4
- Language
- English
- Type
- Conference Paper
- Abstract
- The paper proposes a novel HD-to-UHD video conversion method which analyzes the characteristics of the input HD video and converts it into a high quality UHD video. The proposed method segments the input HD video into plural video shots and analyzes the contrast and edge/texture characteristics of each shot. Based on the analysis, appropriate pre/post filters applied on the upscaled video shots. The upscaling method used here is a self-similarity based super resolution (SSSR) which effectively reconstructs the high-frequency signals of the upscaled video by referring to the local self-similar regions. The contrast characteristics metric is calculated using the intensity histogram of the frames. If the contrast metric is below a certain threshold, the contrast enhancement filter is applied as a pre-process. The edge/texture characteristics metric is calculated by using the density of the dominant edge/texture regions in the frames. If the edge/texture metric is below a certain threshold, the high frequency signals reconstructed from SSSR are boosted as a post-process to increase the sharpness of the shot. Experimental results show that the proposed method can produce a video with improved subjective quality.
- KSP Keywords
- Contrast Enhancement(CE), Conversion method, High frequency(HF), Intensity histogram, Post process, Subjective quality, Super resolution, Texture characteristics, UHD video, Upscaling method, Video conversion