02_自己学习Retinex.ppt

Retinex Image Enhancement Techniques --- Algorithm, Application and Advantages Introduction Why called Retinex? An method bridging the gap between images and the human observation of scenes. Origin of Retinex Proposed by Edwin Land1 in 1986 A model of lightness and color perception of human vision No theoretical but experimentally proved Retinex An automatic imaging process Independent of variations in the scene What could Retinex do? Depending on the circumstances, Retinex could achieve Sharpening Compensation for the blurring introduced by image formation process Color constancy processing Improve consistency of output as illumination changes dynamic range compression Development of Retinex techniques Single Scale Retinex (SSR) Multi-Scale Retinex (MSR) Multi-Scale Retinex with Color Restoration (MSRCR) Multi-Scale Retinex with canonical gain/offset Single Scale Retinex (SSR) Algorithm Ii(x,y): the image distribution in the ith spectral band Ri(x,y): retinex output Gaussian function: F(x,y)=Ke-(x2+y2)/c2 K determined by: C is the Gaussian surround space constant SSR result comparison with different gaussian constant I SSR result comparison with different gaussian constant II Properties of Retinex Multi-Scale Retinex (MSR) Algorithm N: number of scales, ωn: weight associated with the nth scale Empirical value: N=3, ωn=1/3, C = 15, 80 and 250 correspondingly for each scale in Fn Better than SSR in balance of dynamic compression and color rendition Comparison of SSR and MSR Improvements on MSR -- Color Restoration MSR is good enough for gray pictures But not desirable for color pictures RGB proportion out of balance IR(x,y):IG(x,y):IB(x,y) == Solutions Multi-Scale Retinex with Color Restoration (MSRCR) Multi-scale Retinex with color Restoration (MSRCR) Algorithm Further improvements on MSR -- For better contrast  Characteristics of retinex pictures histogram Solutions Canonical gain/offset Canonical: general constants