wolffd@0: function varargout = mirstd(f,varargin)
wolffd@0: % m = mirstd(f) returns the standard deviation along frames of the feature f
wolffd@0: %
wolffd@0: %   f can be a structure array composed of features. In this case,
wolffd@0: %       m will be structured the same way.
wolffd@0: 
wolffd@0: if isa(f,'mirstruct')
wolffd@0:     data = get(f,'Data');
wolffd@0:     for fi = 1:length(data)
wolffd@0:         data{fi} = mirstd(data{fi});
wolffd@0:     end
wolffd@0:     varargout = {set(f,'Data',data)};
wolffd@0: elseif isstruct(f)
wolffd@0:     fields = fieldnames(f);
wolffd@0:     for i = 1:length(fields)
wolffd@0:         field = fields{i};
wolffd@0:         stat.(field) = mirstd(f.(field));
wolffd@0:     end
wolffd@0:     varargout = {stat};
wolffd@0: else
wolffd@0:     
wolffd@0:         normdiff.key = 'NormDiff';
wolffd@0:         normdiff.type = 'Boolean';
wolffd@0:         normdiff.default = 0;
wolffd@0:     specif.option.normdiff = normdiff;
wolffd@0: 
wolffd@0:     specif.nochunk = 1;
wolffd@0:     
wolffd@0:     varargout = mirfunction(@mirstd,f,varargin,nargout,specif,@init,@main);
wolffd@0: end
wolffd@0: 
wolffd@0: 
wolffd@0: function [x type] = init(x,option)
wolffd@0: type = '';
wolffd@0: 
wolffd@0: 
wolffd@0: function m = main(f,option,postoption)
wolffd@0: if iscell(f)
wolffd@0:     f = f{1};
wolffd@0: end
wolffd@0: if isa(f,'mirhisto')
wolffd@0:     warning('WARNING IN MIRSTD: histograms are not taken into consideration yet.')
wolffd@0:     m = struct;
wolffd@0:     return
wolffd@0: end
wolffd@0: fp = get(f,'FramePos');
wolffd@0: ti = get(f,'Title');
wolffd@0: d = get(f,'Data');
wolffd@0: l = length(d);
wolffd@0: for i = 1:l
wolffd@0:     if iscell(d{i})
wolffd@0:         if length(d{i}) > 1
wolffd@0:             error('ERROR IN MIRSTD: segmented data not accepted yet.');
wolffd@0:         else
wolffd@0:             dd = d{i}{1};
wolffd@0:         end
wolffd@0:     else
wolffd@0:         dd = d{i};
wolffd@0:     end
wolffd@0:     if iscell(dd)
wolffd@0:         m{i} = {zeros(1,length(dd))};
wolffd@0:         for j = 1:length(dd)
wolffd@0:             m{i}{1}(j) = std(dd{j});
wolffd@0:         end
wolffd@0:     elseif size(dd,2) < 2
wolffd@0:         nonan = find(not(isnan(dd)));
wolffd@0:         dn = dd(nonan);
wolffd@0:         if option.normdiff
wolffd@0:             m{i}{1} = norm(diff(dn,2));
wolffd@0:         else
wolffd@0:             m{i}{1} = std(dn,0,2);
wolffd@0:         end
wolffd@0:     else
wolffd@0:         dd = mean(dd,4);
wolffd@0:         m{i} = {NaN(size(dd,1),1,size(dd,3))};
wolffd@0:         for k = 1:size(dd,1)
wolffd@0:             for l = 1:size(dd,3)
wolffd@0:                 dk = dd(k,:,l);
wolffd@0:                 nonan = find(not(isnan(dk)));
wolffd@0:                 if not(isempty(nonan))
wolffd@0:                     dn = dk(nonan);
wolffd@0:                     if option.normdiff
wolffd@0:                         m{i}{1}(k,1,l) = norm(diff(dn,2));
wolffd@0:                     else
wolffd@0:                         m{i}{1}(k,1,l) = std(dn,0,2);
wolffd@0:                     end
wolffd@0:                 end
wolffd@0:             end
wolffd@0:         end
wolffd@0:     end
wolffd@0: end
wolffd@0: m = mirscalar(f,'Data',m,'Title',['Standard deviation of ',ti]);