LCIS ABoundary Hierarchy For DetailPreserving Contrast LCIS为保细节层次对比边界.pptVIP

LCIS: A Boundary Hierarchy For Detail-Preserving Contrast Reduction Jack Tumblin and Greg Turk Georgia Institute of Technology SIGGRAPH 1999 Presented by Rob Glaubius Motivation Detail visible almost everywhere in a scene Difficult to capture rich detail in high-contrast scenes CRT contrast: 100:1 Target scene contrast: ~100,000:1 Motivation Simple scene intensity adjustment Id = F(m·Is?) Id: display intensity Is: scene intensity m: scale factor ? : compression/expansion term F: enforces boundary conditions LCIS - A Preview “Mathematically mimic a well-known artistic technique for rendering high contrast scenes” Coarse-to-fine rendering of boundaries and shading LCIS - A Preview Low Curvature Image Simplifier Hierarchy of sharp boundaries and smooth shadings Goal - low contrast, highly detailed images LCIS vs. Linear Filter Hierarchies Anisotropic Diffusion Treat intensity as heat fluid Temperature wants to flow from hot to cold It = ?·(C(x,y,t) ?I) It : derivative of temperature change w.r.t. time C : Conductivity Constant conductivity ? repeated convolution with a Gaussian filter (isotropic diffusion) Anisotropic Diffusion, cont’d Conductivity depends on image - as local “edginess” increases, conductivity decreases C(x,y,t) = g(||?I||) where g(m) = (1+(m/K)2)-1 K is a conductance threshold for m Anisotropic Diffusion Illustrated LCIS vs. Anisotropic Diffusion LCIS - Theory 3rd order derivatives instead of 2nd order Equalize curvature rather than intensity It(x,y,t) = ?·(C(x,y,t)F(x,y,t)) F: motive force from high to low curvature F = (Ixxx + Iyyx, Ixxy + Iyyy) C: Conductivity C(x,y,t) = g(0.5(I2xx + I2yy) + I2xy) LCIS - Implementation Discrete images, so quantities are approximate, based on 4-connected neighbors and a constant time step LCIS Hierarchy * * Convert (Rin,Gin,Bin) ? ? ? LCIS K0 = 0 LCIS K1 LCIS K2 LCIS K3 + + + (Rout,Gout,Bout) exp() ? ? ? ? ? wcolor w0 w1 w2 w3 log(L) log(R/L,G/L,B/L) * * *