windows下可啟動並執行mat轉xml，VOC-release4.01 DPM訓練的model（mat）轉為OpenCV latentsvm可以載入的model（xml）

轉換後的single component person model檢測效果不錯，而VOC2007 2009中的three-component person model的檢測效果則非常差，可能是OpenCV對混合模型的載入不太支援。（我只關心person的檢測，所以只測試了person的model轉換後的效果）

% jelly 2013-08-12
% Convert *.mat format model in the source example in % Discriminatively Trained Deformable Part Models "voc-release4.01"% to opencv's latentSVM detect input format *.xml
function MAT2XMLmodel_401(matmodel, xmlfile)
matmodel = 'INRIA/inriaperson_final';xmlfile = 'INRIA/inriaperson_final.xml';
load(matmodel);fid = fopen(xmlfile, 'w');
fprintf(fid, '<Model>\n');%%%擷取組件數ncom = length(model.rules{model.start});fprintf(fid, '\t<!-- Number of components -->\n');fprintf(fid, '\t<NumComponents>%d</NumComponents>\n', ncom);%擷取特徵維數，固定值31維。model中沒有記錄nfeature = 31;fprintf(fid, '\t<!-- Number of features -->\n');fprintf(fid, '\t<P>%d</P>\n', nfeature);%Score threshold=model.threshfprintf(fid, '\t<!-- Score threshold -->\n');fprintf(fid, '\t<ScoreThreshold>%.16f</ScoreThreshold>\n', model.thresh);layer = 1;%對於每一個組件分別擷取它的root filter、part filter、deformation filterfor icom = 1:ncom    fprintf(fid, '\t<Component>\n');            fprintf(fid, '\t\t<!-- Root filter description -->\n');        fprintf(fid, '\t\t<RootFilter>\n');                    % attention: X,Y swap            rhs = model.rules{model.start}(icom).rhs;            % assume the root filter is first on the rhs of the start rules            if model.symbols(rhs(1)).type == 'T'              % handle case where there's no deformation model for the root              root = model.symbols(rhs(1)).filter;            else              % handle case where there is a deformation model for the root              root = model.symbols(model.rules{rhs(1)}(layer).rhs).filter;            end                        filternum = root;            sizeX = model.filters(filternum).size(2);            sizeY = model.filters(filternum).size(1);            fprintf(fid, '\t\t\t<!-- Dimensions -->\n');             fprintf(fid, '\t\t\t<sizeX>%d</sizeX>\n', sizeX);             fprintf(fid, '\t\t\t<sizeY>%d</sizeY>\n', sizeY);                    fprintf(fid, '\t\t\t<!-- Weights (binary representation) -->\n');            fprintf(fid, '\t\t\t<Weights>');            for iY = 1:sizeY                for iX = 1:sizeX                    % original mat has 32 which is larger than nfeature=31 by 1                    fwrite(fid, model.filters(filternum).w(iY,iX,1:nfeature), 'double'); % need verify                end            end            fprintf(fid, '\t\t\t</Weights>\n');                              fprintf(fid, '\t\t\t<!-- Linear term in score function -->\n');            fprintf(fid, '\t\t\t<LinearTerm>%.16f</LinearTerm>\n', ...  % need verify                model.rules{model.start}(icom).offset.w);                    fprintf(fid, '\t\t</RootFilter>\n');                fprintf(fid, '\t\t<!-- Part filters description -->\n');        fprintf(fid, '\t\t<PartFilters>\n');        %在每個component內，擷取part filter的個數，並擷取每個part的參數        npart = length(model.rules{model.start}(icom).rhs) -1 ;        fprintf(fid, '\t\t\t<NumPartFilters>%d</NumPartFilters>\n', npart);        %%擷取每個part的相關參數[dx,dy,ds]和penalty[dx dy dxx dyy]        for ipart = 2: npart+1            fprintf(fid, '\t\t\t<!-- Part filter ? description -->\n');            fprintf(fid, '\t\t\t<PartFilter>\n');                        irule = model.rules{model.start}(icom).rhs(ipart);            filternum = model.symbols(model.rules{irule}.rhs).filter;            sizeX = model.filters(filternum).size(2);            sizeY = model.filters(filternum).size(1);            fprintf(fid, '\t\t\t\t<sizeX>%d</sizeX>\n', sizeX);            fprintf(fid, '\t\t\t\t<sizeY>%d</sizeY>\n', sizeY);            fprintf(fid, '\t\t\t\t<!-- Weights (binary representation) -->\n');            fprintf(fid, '\t\t\t\t<Weights>');            for iY = 1:sizeY                for iX = 1:sizeX                    % original mat has 32 which is larger than nfeature=31 by 1                    fwrite(fid, model.filters(filternum).w(iY,iX,1:nfeature), 'double'); % need verify                end            end            fprintf(fid, '\t\t\t\t</Weights>\n');                        fprintf(fid, '\t\t\t\t<!-- Part filter offset -->\n');                        fprintf(fid, '\t\t\t\t<V>\n');                        fprintf(fid, '\t\t\t\t\t<Vx>%d</Vx>\n',model.rules{model.start}(icom).anchor{ipart}(1)+1); %[dx,dy,ds]            fprintf(fid, '\t\t\t\t\t<Vy>%d</Vy>\n',model.rules{model.start}(icom).anchor{ipart}(2)+1);                        fprintf(fid, '\t\t\t\t</V>\n');                                    fprintf(fid, '\t\t\t\t<!-- Quadratic penalty function coefficients -->\n');                        fprintf(fid, '\t\t\t\t<Penalty>\n');            fprintf(fid, '\t\t\t\t\t<dx>%.16f</dx>\n',model.rules{irule}.def.w(2));             fprintf(fid, '\t\t\t\t\t<dy>%.16f</dy>\n',model.rules{irule}.def.w(4));             fprintf(fid, '\t\t\t\t\t<dxx>%.16f</dxx>\n',model.rules{irule}.def.w(1));             fprintf(fid, '\t\t\t\t\t<dyy>%.16f</dyy>\n',model.rules{irule}.def.w(3));                         fprintf(fid, '\t\t\t\t</Penalty>\n');                            fprintf(fid, '\t\t\t</PartFilter>\n');        end            fprintf(fid, '\t\t</PartFilters>\n');    fprintf(fid, '\t</Component>\n');endfprintf(fid, '</Model>\n');fclose(fid);
end