作者: Ayan Chakrabarti , William T. Freeman , Todd Zickler
DOI: 10.1109/ICCPHOT.2014.6831801
关键词:
摘要: Digital color cameras make sub-sampled measurements of at alternating pixel locations, and then “demosaick” these to create full images by up-sampling. This allows traditional with restricted processing hardware produce from a single shot, but it requires blocking majority the incident light is prone aliasing artifacts. In this paper, we introduce computational approach photography, where sampling pattern reconstruction process are co-designed enhance sharpness photographic speed. The made predominantly panchromatic, thus avoiding excessive loss high spatial-frequency intensity variations. Color sampled very sparse set locations propagated throughout image guidance un-aliased luminance channel. Experimental results show that often leads significant reductions in noise artifacts, especially low-light conditions.