基于Mumford-Shah模型的水平集图像分割算法

function K = CURVATURE(f,diff_scheme) % CURVATURE computes curvature % Author: Chunming Li, all rights reserved. % Email: li_chunming@hotmail.com % URL: http://www.engr.uconn.edu/~cmli/ epsilon=1e-10; if strcmp(diff_scheme, 'fcb') [fx,fy]=gradient(f); % central difference fx_f = Dx_forward(f); % forward difference ax = fx_f./sqrt(fx_f.^2+ fy.^2+epsilon); axx = Dx_backward(ax); % backward difference fy_f = Dy_forward(f); ay = fy_f./sqrt(fx.^2 + fy_f.^2 + epsilon); ayy = Dy_backward(ay); K = axx + ayy; elseif strcmp(diff_scheme, 'fb') % forward difference followed by a backward difference fx_f = Dx_forward(f); fy_f = Dy_forward(f); ax = fx_f./sqrt(fx_f.^2+ fy_f.^2+epsilon); ay = fy_f./sqrt(fx_f.^2 + fy_f.^2 + epsilon); axx = Dx_backward(ax); ayy = Dy_backward(ay); K = axx + ayy; elseif strcmp(diff_scheme, 'bf') % forward difference followed by a backward difference fx_f = Dx_backward(f); fy_f = Dy_backward(f); ax = fx_f./sqrt(fx_f.^2+ fy_f.^2+epsilon); ay = fy_f./sqrt(fx_f.^2 + fy_f.^2 + epsilon); axx = Dx_forward(ax); ayy = Dy_forward(ay); K = axx + ayy; elseif strcmp(diff_scheme, 'cc') % central difference followed by a central difference [fx, fy]= gradient(f); % central difference ax = fx./sqrt(fx.^2+ fy.^2+epsilon); ay = fy./sqrt(fx.^2 + fy.^2 + epsilon); [axx, axy] = gradient(ax); % central difference [ayx, ayy] = gradient(ay); K = axx + ayy; else disp('Wrong difference scheme: CURVATURE.m'); return; end