Replace ~ with dummy variables to be compatible to Matlab <R2009b.

Tested, working on Ubuntu10.10 32bit, opencv 2.1, Matlab R2009a. 
Runs on octave 3.4. Picture sequence is shown, tracking (blue dots) unsuccessful, process interruption with keypress doesn't work, Tracking output mostly NaN.

tld/tldDetection.m

@@ -29,7 +29,7 @@
 idx_dt = find(tld.tmp.conf > tld.model.num_trees*tld.model.thr_fern); % get indexes of bounding boxes that passed throu the Ensemble Classifier
 
 if length(idx_dt) > 100 % speedup: if there are more than 100 detections, pict 100 of the most confident only
-    [~,sIdx] = sort(tld.tmp.conf(idx_dt),'descend');
+    [dummy8,sIdx] = sort(tld.tmp.conf(idx_dt),'descend');
     idx_dt = idx_dt(sIdx(1:100));
 end
 

tld/tldExample.m

@@ -24,7 +24,7 @@
 opt.source = tldInitSource(opt.source); % select data source, camera/directory
 
 figure(2); set(2,'KeyPressFcn', @handleKey); % open figure for display of results
-finish = 0; function handleKey(~,~), finish = 1; end % by pressing any key, the process will exit
+finish = 0; function handleKey(dummy1,dummy2), finish = 1; end % by pressing any key, the process will exit
 
 while 1
     source = tldInitFirstFrame(tld,opt.source,opt.model.min_win); % get initial bounding box, return 'empty' if bounding box is too small

tld/tldGeneratePositiveData.m

@@ -21,13 +21,13 @@
 pEx  = [];%zeros(prod(tld.patchsize),numWarps);
 
 % Get closest bbox
-[~,idxP] = max(overlap);
+[dummy3,idxP] = max(overlap);
 bbP0 =  tld.grid(1:4,idxP);
 
 % Get overlapping bboxes
 idxP = find(overlap > 0.6);
 if length(idxP) > p_par.num_closest
-    [~,sIdx] = sort(overlap(idxP),'descend');    
+    [dummy4,sIdx] = sort(overlap(idxP),'descend');    
     idxP = idxP(sIdx(1:p_par.num_closest));
 end
 bbP  = tld.grid(:,idxP);

tld/tldLearning.m

@@ -23,7 +23,7 @@
 % Check consistency -------------------------------------------------------
 
 pPatt  = tldGetPattern(img,bb,tld.model.patchsize); % get current patch
-[pConf1,~,pIsin] = tldNN(pPatt,tld); % measure similarity to model
+[pConf1,dummy9,pIsin] = tldNN(pPatt,tld); % measure similarity to model
 
 if pConf1 < 0.5, disp('Fast change.'); tld.valid(I) = 0; return; end % too fast change of appearance
 if var(pPatt) < tld.var, disp('Low variance.'); tld.valid(I) = 0; return; end % too low variance of the patch

tld/tldNN.m

@@ -44,7 +44,7 @@
 
     % set isin
     if any(nccP > tld.model.ncc_thesame), isin(1,i) = 1;  end % IF the query patch is highly correlated with any positive patch in the model THEN it is considered to be one of them
-    [~,isin(2,i)] = max(nccP); % get the index of the maximall correlated positive patch
+    [dummy6,isin(2,i)] = max(nccP); % get the index of the maximall correlated positive patch
     if any(nccN > tld.model.ncc_thesame), isin(3,i) = 1;  end % IF the query patch is highly correlated with any negative patch in the model THEN it is considered to be one of them
 
     % measure Relative Similarity

tld/tldTracking.m

@@ -41,7 +41,7 @@
 
 % estimate confidence and validity
 patchJ   = tldGetPattern(tld.img{J},BB2,tld.model.patchsize); % sample patch in current image
-[~,Conf] = tldNN(patchJ,tld); % estimate its Conservative Similarity (considering 50% of positive patches only)
+[dummy7,Conf] = tldNN(patchJ,tld); % estimate its Conservative Similarity (considering 50% of positive patches only)
 
 % Validity
 Valid    = tld.valid(I); % copy validity from previous frame

tld/tldTrainNN.m

@@ -33,7 +33,7 @@
 for k = 1:1 % Bootstrap
    for i = 1:length(y)
 
-       [conf1,~,isin] = tldNN(x(:,i),tld); % measure Relative similarity
+       [conf1,dummy5,isin] = tldNN(x(:,i),tld); % measure Relative similarity
 
        % Positive
        if y(i) == 1 && conf1 <= tld.model.thr_nn % 0.65