matlab-(含教程)基于双目视觉的人脸三维建模matlab仿真

function [Iout]=ut_edge(varargin) %ut_edge Marr-Hilderth and Canny edge detectors % % Iout = UT_EDGE(Iin) finds the possible edges in image Iin using an % implementation of the Canny edge detector. The derivatives are % calculated from a 2-D Gaussian function implemented with UT_GAUSS. % The localization is done by finding the zero-crossings of the second % directional derivatives (instead of directional maximum suppression % such as implemented by EDGE from the image processing toolbox). The % sigma (standard deviation) of the Gaussian filtering is set to 2. % Iout is a 2-dimensional array with the same size as Iin. The no-edge % pixels are indicated in Iout as zeroes. Edge pixels carries the edge % strength as grey level. The edge strength is scaled between 0 and 1. % % Iout = UT_EDGE(Iin,...,OPTION,OPTIONVAL,...) finds the possible edges % with some options switched on, and with parameter(s) set to OPTIONVAL. % The following options are available: % % 'marrhildreth' or 'm': Perform Marr-Hildreth operation, i.e. Laplacian % of Gaussian, rather than Canny operation. % 'canny' or 'c': Perform Canny operation (default). % 'sigma' or 's': The width (standard deviation) of the Gaussian % filtering is set to OPTIONVAL (default: 2) % 'threshold' or 't': Return a binary (logical) edge map instead of edge % strengths. The edges are detected by % thresholding it at OPTIONVAL. If OPTIONVAL is % omitted, the default of the threshold is 0.1. % 'hysteresis' or 'h': Return a binary (logical) edge map instead of edge % strengths. The edge are detected by hysteresis % thresholding. Two thresholds are applied. The % edges detected with the higher thresholds form % the seed that is propagated in the edge % segments obtained by thresholding with the % lower threshold. OPTIONVAL is a two-element % vector. The first element specifies the % fraction of pixels that will be assigned to % candidate edges (detected with the lower % threshold). The second element is the fraction % of pixels that will be assigned to seed edges % (detected with the higher threshold). If % OPTIONVAL is omitted, the default of [0.1 0.03] % is assumed. If OPTIONVAL is a scalar, then the % default assignment to the second element is % 0.3*OPTIONVAL. % % Copyright: F. van der Heijden, F.vanderHeijden@utwente.nl % Laboratory for Measurements and Instrumentation % University of Twente, the Netherlands % Version 1.3, date: 21-02-2018 % % See also UT_GAUSS, EDGE. [Iin,sigma,operator,threshold_method,thres_parm]=ParseInputs(varargin{:}); grad=(ut_gauss(Iin,sigma,1,0).^2+ut_gauss(Iin,sigma,0,1).^2).^0.5; grad=grad/max(grad(:)); if strcmp(operator,'marrhildreth') lap=ut_gauss(Iin,sigma,2,0)+ut_gauss(Iin,sigma,0,2); zerocross=ut_levelx(lap,0.0); Iout=grad.*zerocross; end if strcmp(operator,'canny') SDGD=ut_gauss(Iin,sigma,1,0).^2.*ut_gauss(Iin,sigma,2,0)+ ... 2*ut_gauss(Iin,sigma,1,0).*ut_gauss(Iin,sigma,0,1).*ut_gauss(Iin,sigma,1,1)+... ut_gauss(Iin,sigma,0,1).^2.*ut_gauss(Iin,sigma,0,2); zerocross=ut_levelx(SDGD,0.0); Iout=grad.*zerocross; end if strcmp(threshold_method,'threshold') Iout = logical(Iout>thres_parm(1)); end if strcmp(threshold_method,'hysteresis') Iout_sorted = sort(Iout(:)); thres_parm(1) = Iout_sorted(max(floor((1-thres_parm(1))*numel(Iout)),1)); thres_parm(2) = Iout_sorted(max(floor((1-thres_parm(2))*numel(Iout)),1)); Iout_mask = Iout>thres_parm(1); Iout_marker = Iout>thres_parm(2); Iout=imreconstruct(Iout_marker,Iout_mask); end %---------------------------------------------------------------------- % Subfunction ParseInputs %---------------------------------------------------------------------- function [Iin,Sigma,operator,threshold_method,thres_parm] = ParseInputs(varargin) narginchk(1,30); Iin = double(varargin{1}); if ndims(Iin)~=2 error('Iin must be a 2-dimensional array'); end %defaults Sigma=2; operator = 'canny'; threshold_method = 'none'; thres_parm = [0.1, 0.03]; options = {'canny','c','marrhildreth','m','threshold','t','hysteresis','h','sigma','s'}; if nargin>1 for k=2:length(varargin) if ischar(varargin{k}) string = lower(varargin{k}); j = strmatch(string,options,'exact'); if isempty(j) error(['Invalid input string: ''' varargin{k} '''.']); end switch string case {'canny','c','marrhildreth','m'} if strcmp(string,'m'), string = 'marrhildreth'; end; if strcmp(string,'c'), string = 'canny'; end; operator = string; case {'sigma','s'} if nargin>k & isnumeric(varargin{k+1}) Sigma = varargin{k+1}; k=k+1; end case {'threshold','t'} threshold_method = 'threshold'; thres_parm = [0.1 0.03]; if nargin>k & isnumeric(varargin{k+1}) thres_parm(1) = varargin{k+1}; k=k+1; end case {'hysteresis','h'} threshold_method = 'hysteresis'; thres_parm = [0.1 0.03]; if nargin>k & isnumeric(varargin{k+1}) thres_parm = varargin{k+1}; if length(thres_parm)==1 thres_parm(2) = 0.3*thres_parm(1); end k=k+1; end end end end end if Sigma<0 error('Sigma must be positive'); end if strcmp(threshold_method , 'hysteresis') if thres_parm(1)<0 | thres_parm(1)>1 | thres_parm(2)<0 | thres_parm(2)>1 error('hysteresis: threshold parameters out of range'); end if thres_parm(2)>thres_parm(1) error('hysteresis: the second threshold must be less than the first'); end end