Improve the Schizont detection performance.

diff --git a/src/DetectionOfParasites.m b/src/DetectionOfParasites.m
index bdb2a51..9dca5c4 100644 (file)
--- a/src/DetectionOfParasites.m
+++ b/src/DetectionOfParasites.m
@@ -13,7 +13,7 @@ function [THS, c] = DetectionOfParasites(imgRGB)
 
     imgH = imgHSV(:, :, 1) * 255;
     % We shift the hue to have the greatest value for the nuclei.
-    hueShiftValue = 175; % 100
+    hueShiftValue = 175;
     imgH = uint8(mod(255 - imgH + hueShiftValue, 256));
     imwrite(imgH, '../output/imgH.png')
 
@@ -23,9 +23,6 @@ function [THS, c] = DetectionOfParasites(imgRGB)
     imgV = uint8(255 - imgHSV(:, :, 3) * 255);
     imwrite(imgV, '../output/imgV.png')
 
-    % imgV = uint8(imgHSV(:, :, 3) * 255); % The value component isn't used.
-    % imgH = 255 - imgV; % We use the value component instead of the hue one (just for testing)
-
     % p. 136: median filter to smooth the noise and 
     % area closing to enhance the bright objects and make flatter, darker and cleaner the image background.
     imgFiltered{1} = mmareaclose(medfilt2(imgH, [5, 5]), 400); % Hue.
@@ -52,21 +49,21 @@ function [THS, c] = DetectionOfParasites(imgRGB)
 
     sizeDistribution = zeros(redCellMaxSize, 1);
     parfor k = 1:redCellMaxSize 
-        SE = mmsedisk(k, '2D', 'OCTAGON');
+        SE = mmsedisk(k, '2D', 'OCTAGON'); % TODO : tester avec 'euclidean'
         imgOpened = mmopen(imgFiltered{2}, SE);
         A = funVolume(imgOpened);
         N = 1 - A / volImg;
         sizeDistribution(k) = N;
     end
 
+    % TODO : voir diff
     patternSpectrum = zeros(redCellMaxSize - 1, 1);
     for k = 1:redCellMaxSize - 1
         patternSpectrum(k) = abs(sizeDistribution(k + 1) - sizeDistribution(k));
     end
 
-    % The paper choose the biggest red cell size among the possible red cell sizes. (FIXME)
+    % The paper chooses the biggest red cell size among the possible red cell sizes. (FIXME)
     [m, c] = max(patternSpectrum);
-    % c = c + 3;
     nucleiRadius = 5; % Find a way to extract this information from the pattern spectrum histogram. (FIXME)
 
     bar(patternSpectrum);

diff --git a/src/DetectionOfSchizonts.m b/src/DetectionOfSchizonts.m
index 4d94f08..e16327d 100644 (file)
--- a/src/DetectionOfSchizonts.m
+++ b/src/DetectionOfSchizonts.m
@@ -1,4 +1,6 @@
 % Inputs:
+%  THS:
+%  c: (TODO: should be the average red cell size)
 function [schizonts] = DetectionOfSchizonts(THS, c)    
 
     redCellsSE = mmsedisk(c, '2D', 'OCTAGON');    
@@ -25,14 +27,19 @@ function [schizonts] = DetectionOfSchizonts(THS, c)
     %% Check the Hausdorff distance between each set combination by dilation.
     for i = 1:length(sets)
         setA = sets{i};
+        if isempty(setA)
+            continue
+        end
         dilatedSetA = mmdil(setA, redCellsSE);
         for j = 1:length(sets)
             setB = sets{j};
-            if setB == setA
+            if isequal(setB, setA) || isempty(setB)
                 continue
             end
-            if (dilatedSetA & setB) == setB %#ok<AND2> % If B is a subset of A -> there are both in a Hausdorff distance of 2*c.
+            if (dilatedSetA & setB) == setB % If B is a subset of A -> there are both in a Hausdorff distance of 2*c.
                 schizonts = schizonts | setA | setB;
+                sets{i} = [];
+                sets{j} = [];
             end
         end
     end    

diff --git a/src/SegmentationOfRedCells.m b/src/SegmentationOfRedCells.m
index e69de29..fe6292b 100644 (file)
--- a/src/SegmentationOfRedCells.m
+++ b/src/SegmentationOfRedCells.m
@@ -0,0 +1 @@
+function 
\ No newline at end of file