Image enhancement based on intrascale dependencies of the second generation curvelet transform
Hongxia Hao, Fang Liu, Licheng Jiao · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2009
This paper presents a strong noise image enhancement method based on intrascale dependencies of the second generation curvelet transform. Observing that the immediate four neighbor coefficients bear the most important dependencies, we use spatial clustering property of the intrascale neighbor coefficients to separate noise and signal of interest, and to deal with them differently, i.e. to suppress noise and strengthen edges. Comparing our approach with Starck's enhancement model (Starck et al., 2003), we experimentally find that for high noise level images, our method outperforms the starck's system in noise suppression and signal strengthening and produces better enhancement results.