Expe on synthetic images

Article/Fig_method/algoAuto.tex

@@ -19,7 +19,7 @@
   \SetKwData{Begin}{Start}
   \SetKwData{End}{End}
   
-  \Input{Image \image, width \iwidth, height \iheight, resolution \resol}
+  \Input{Image \image, width \iwidth, height \iheight, sweeping step \resol}
   \Output{\textit{\bslist} $\rightarrow$ list of detected blurred segments}
   \BlankLine
   \bslist $\leftarrow$ \nullset\;
@@ -41,5 +41,5 @@
     $y \leftarrow y~+$ \resol\;
   }
   
-  \caption{AutoDetect: finds all blurred segments in the image.}
+  \caption{AutoDetect: finds all the blurred segments in the image.}
 \end{algorithm}

Article/Fig_method/algoMulti.tex

@@ -17,14 +17,14 @@
   
   \SetKwFunction{locmax}{ComputeAndSortGradientLocalMax}
   \SetKwFunction{sortgradmax}{SortByGradientMagnitude}
+  \SetKwFunction{updatemask}{UpdateOccupancyMask}
   \SetKwFunction{taille}{Size}
   \SetKwFunction{cardinal}{Card}
 
   \SetKwData{Begin}{Start}
   \SetKwData{End}{End}
   
-  \Input{Stroke points \pta, \ptb}
-  \Input{Occupancy mask \mask}
+  \Input{Stroke points \pta, \ptb, occupancy mask \mask}
   \Output{\textit{\bslist} $\rightarrow$ list of detected blurred segments}
   \BlankLine
   \bslist $\leftarrow$ \nullset\;
@@ -32,11 +32,9 @@
   \BlankLine
   \For{$i \leftarrow 0$ \KwTo \taille(\lm)}{
     \bseg $\leftarrow$ detect (\lm[i], \ortho, \eps, \mask)\;
-    \For{$j \leftarrow 0$ \KwTo \cardinal (\bseg)}{
-      \mask $\leftarrow$ \bseg[j]\;
-  }
+    \updatemask (\mask, \bseg)\;
     \bslist $\leftarrow$ \bseg\;
   }
   
-  \caption{MultiDetect: finds all blurred segments from input selection.}
+  \caption{MultiDetect: finds all the selected blurred segments.}
 \end{algorithm}

Article/Fig_synth/stats.txt

+  RESULTS FOR THE OLD DETECTOR
+56.515 (pm 16.826) segments searches (local min) / image
+21.223 (pm 5.31578) provided segments / image
+10.898 (pm 1.88339) provided long segments / image
+1.15 (pm 1.79798) undetected segments per image
+96.8287 (pm 0.608812) % of points found
+0.0154251 (pm 0.0252677) % of points found more than once (redetections)
+6.92689 (pm 1.0261) % false points produced
+Width difference : 2.08036 (0.266341) per matched segment
+Absolute width difference : 2.10178 (0.264193) per matched segment
+Angle difference : -0.00233791 (0.675674) degrees per matched segment
+Absolute angle difference : 0.872721 (0.780913) per matched segment
+Absolute long edge angle difference : 0.400459 (0.627538) per matched segment
+  RESULTS FOR THE NEW DETECTOR
+61.405 (pm 15.4125) segments searches (local min) / image
+27.334 (pm 6.37535) provided segments / image
+9.03 (pm 2.15261) provided long segments / image
+0.281 (pm 0.562455) undetected segments per image
+96.8063 (pm 0.615733) % of points found
+0.011969 (pm 0.0195106) % of points found more than once (redetections)
+6.5253 (pm 0.94361) % false points produced
+Width difference : 1.90539 (0.236267) per matched segment
+Absolute width difference : 1.94935 (0.230798) per matched segment
+Angle difference : 0.00697535 (0.724535) degrees per matched segment
+Absolute angle difference : 0.916683 (0.754317) per matched segment
+Long edge absolute angle difference : 0.336041 (0.595828) per matched segment

Article/Fig_synth/statsTable.tex

+\begin{tabular}{|l||rcl|rcl|}
+\hline
+Number of input segments per image
+& 21.22 & $\pm$ & 5.32 & 27.33 & $\pm$ & 6.38 \\
+Number of long input segments per image
+& 10.90 & $\pm$ & 1.88 & 9.03 & $\pm$ & 2.15 \\
+Number of undetected segments per image
+& 1.15 & $\pm$ & 1.80 & 0.28 & $\pm$ & 0.56 \\
+Percentage of true segment area found
+& 96.83 & $\pm$ & 0.61 & 96.81 & $\pm$ & 0.62 \\
+Percentage of false segment area found
+& 6.9 & $\pm$ & 1.0 & 6.52 & $\pm$ & 0.94 \\
+Width difference (in pixels) to input segment
+& 2.08 & $\pm$ & 0.27 & 1.91 & $\pm$ & 0.24 \\
+Angle difference (in degrees) to input segment
+& 0.87 & $\pm$ & 0.78 & 0.92 & $\pm$ & 0.75 \\
+Long segment angle difference (in degrees)
+& 0.40 & $\pm$ & 0.63 & 0.34 & $\pm$ & 0.60 \\
+\hline
+\end{tabular}

Article/biblio.bib

@@ -56,7 +56,7 @@
   number = "6",
   pages = "1675--1684",
   year = "2007",
-  doi = "10.1016/j.patcog.2006.10.005"
+  optdoi = "10.1016/j.patcog.2006.10.005"
 }
 
 
@@ -81,7 +81,7 @@
   year = {2017},
   number = {2},
   pages = {35-54},
-  doi = {10.1016/j.patcog.2006.10.005}
+  optdoi = {10.1016/j.patcog.2006.10.005}
 }
 
 
@@ -124,7 +124,7 @@
   optmonth = apr,
   year = {2010},
   pages = {722--732},
-  doi = {10.1109/TPAMI.2008.300}
+  optdoi = {10.1109/TPAMI.2008.300}
 }
 
 
@@ -147,7 +147,7 @@
   publisher = {IEEE},
   year = {2015},
   pages = {507--511},
-  doi = {10.1109/ICIP.2015.7350850}
+  optdoi = {10.1109/ICIP.2015.7350850}
 }
 
 
@@ -161,5 +161,5 @@
   number = {01},
   year = {2012},
   pages = {1255002},
-  doi = {10.1142/S0218001412550026}
+  optdoi = {10.1142/S0218001412550026}
 }

Article/expe.tex

@@ -9,6 +9,10 @@ detection followed by two refinement steps) is integrated as an option
 into the code of the new detector, so that both methods rely on the same
 optimized basic routines.
 
+\input{expeSynthese}
+
+\input{expeHard}
+
 The first test compares the computation times of both detectors on a
 selection of input strokes (\RefFig{fig:buro}). Results are displayed
 in \RefTab{tab:cmpOldNew}.

Article/expeHard.tex

+
+The second test (\RefFig{fig:hard}) compares the performance of both
+detectors on a quite difficult image with a lot of gradient noise. 
+The new detector provides less outliers and misaligned segments, and
+globally more relevant informations to infere the structure of the brick wall.
+\begin{figure}[h]
+%\center
+  \begin{tabular}{
+      c@{\hspace{0.1cm}}c@{\hspace{0.1cm}}c@{\hspace{0.1cm}}}
+    \includegraphics[width=0.32\textwidth]{Fig_method/parpaings.png} &
+    \includegraphics[width=0.32\textwidth]{Fig_hard/hardOld.png} &
+    \includegraphics[width=0.32\textwidth]{Fig_hard/hardNew.png}
+    \begin{picture}(1,1)
+      {\color{dwhite}{
+        \put(-286,4.5){\circle*{8}}
+        \put(-171,4.5){\circle*{8}}
+        \put(-58,4.5){\circle*{8}}
+      }}
+      \put(-288.5,2){a}
+      \put(-173.5,2){b}
+      \put(-60.5,2){c}
+    \end{picture}
+  \end{tabular}
+  \caption{Evaluation on a quite textured image:
+           1) imput image,
+           2) results of the old detector,
+           3) results of the new detector.}
+  \label{fig:hard}
+\end{figure}

Article/expeSynthese.tex

+
+The first test (\RefFig{fig:synth}) compares the performance of both
+detectors on a set of 1000 synthesized images containing 10 randomly
+placed straight segments with random width between 1 and 4 pixels.
+Altough this perfect world context with low gradient noise tends to soften
+the old detector weaknesses, the results of \RefTab{tab:synth} show slightly
+better measurements of width and angle on long segments for the new detector.
+The new detector generates more small segments that degrade the angle
+estimations, but it produces a smaller amount of false detections and
+succeeds in finding most of the segments.
+\begin{figure}[h]
+%\center
+  \begin{tabular}{
+      c@{\hspace{0.1cm}}c@{\hspace{0.1cm}}c@{\hspace{0.1cm}}c@{\hspace{0.1cm}}c}
+    \includegraphics[width=0.19\textwidth]{Fig_synth/statsExample.png} &
+    \includegraphics[width=0.19\textwidth]{Fig_synth/statsoldPoints.png} &
+    \includegraphics[width=0.19\textwidth]{Fig_synth/statsoldBounds.png} &
+    \includegraphics[width=0.19\textwidth]{Fig_synth/statsnewPoints.png} &
+    \includegraphics[width=0.19\textwidth]{Fig_synth/statsnewBounds.png}
+    \begin{picture}(1,1)
+      \put(-310,0){1)}
+      \put(-240,0){2)}
+      \put(-170,0){3)}
+      \put(-100,0){4)}
+      \put(-30,0){5)}
+    \end{picture}
+  \end{tabular}
+  \caption{Evaluation on synthesized images:
+           1) one of the test images,
+           2) output segments points from the old detector and
+           3) their minimal digital straight segments,
+           4) output segments points from the new detector and
+           5) their minimal digital straight segments.}
+  \label{fig:synth}
+\end{figure}
+\begin{table}
+\centering
+\input{Fig_synth/statsTable}
+\caption{Measured performance of both detectors on a set of synthesized images.}
+\label{tab:synth}
+\end{table}

Article/method.tex

@@ -257,6 +257,8 @@ In order to avoid multiple detections of the same edge, an occupancy mask,
 initially empty, collects the dilated points of all the blurred segments,
 so that these points can not be added to another segment.
 
+\input{Fig_method/algoMulti}
+
 First the positions $M_j$ of the prominent local maxima of the gradient
 magnitude found under the stroke are sorted from the highest to the lowest.
 For each of them the main detection process is run with three modifications:
@@ -265,14 +267,13 @@ For each of them the main detection process is run with three modifications:
 $\vec{AB}_\perp$ to the stroke as input to build a static scan of fixed width
 $2~\varepsilon_{ini}$, and $M_j$ is used as start point of the blurred
 segment;
-\item the occupancy mask is filled in with the points of the detected blurred
-segments $\mathcal{B}_j''$ at the end of each successful detection;
+\item the occupancy mask is filled in with the points of the dilated blurred
+segments $\mathcal{B}_j''$ at the end of each successful detection
+(a 21 pixels bowl is used to get the segment dilation);
 \item points marked as occupied are rejected when selecting candidates for the
 blurred segment extension in the fine tracking step.
 \end{enumerate}
 
-\input{Fig_method/algoMulti}
-
 In edge selection mode (\RefFig{fig:edgeDir} b), the multi-detection
 algorithm is executed twice, first in main edge selection mode, then
 in opposite edge selection mode.
@@ -334,41 +335,41 @@ is described in Algorithm 2. A stroke that crosses the whole image, is
 swept in both direction, vertical then horizontal, from the center to
 the borders. At each position, the multi-detection algorithm is run
 to collect all the segments found under the stroke.
+In the present work, the stroke sweeping step $\delta$ is set to 10 pixels.
 
 \input{Fig_method/algoAuto}
 
-\RefFig{fig:evalAuto}b gives an idea of the automatic detection performance.
-In the example of \RefFig{fig:noisy}, hardly perceptible edges are detected
-despite of a quite textured context.
-Unsurpringly the length of the detected edges is linked to the initial
-value of the assigned width, but a large value also augments the rate
-of interfering outliers insertion.
-
-\begin{figure}[h]
-\center
-  \begin{tabular}{c@{\hspace{0.1cm}}c@{\hspace{0.1cm}}c}
-    \includegraphics[width=0.32\textwidth]{Fig_method/parpaings.png} &
-    \includegraphics[width=0.32\textwidth]{Fig_method/parpaings2.png} &
-    \includegraphics[width=0.32\textwidth]{Fig_method/parpaings3.png}
-  \end{tabular}
-  \begin{picture}(1,1)(0,0)
-    {\color{dwhite}{
-      \put(-286,-25.5){\circle*{8}}
-      \put(-171,-25.5){\circle*{8}}
-      \put(-58,-25.5){\circle*{8}}
-    }}
-    \put(-288.5,-28){a}
-    \put(-173.5,-28){b}
-    \put(-60.5,-28){c}
-  \end{picture}
-  \caption{Automatic detection of blurred segments on a textured image.
-           a) the input image,
-           b) automatic detection result with initial assigned width set
-           to 3 pixels,
-           c) automatic detection result with initial assigned width set
-           to 8 pixels.}
-  \label{fig:noisy}
-\end{figure}
+%\RefFig{fig:evalAuto}b gives an idea of the automatic detection performance.
+%In the example of \RefFig{fig:noisy}, hardly perceptible edges are detected
+%despite of a quite textured context.
+%Unsurpringly the length of the detected edges is linked to the initial
+%value of the assigned width, but a large value also augments the rate
+%of interfering outliers insertion.
+%
+%\begin{figure}[h]
+%\center \begin{tabular}{c@{\hspace{0.1cm}}c@{\hspace{0.1cm}}c}
+%    \includegraphics[width=0.32\textwidth]{Fig_method/parpaings.png} &
+%    \includegraphics[width=0.32\textwidth]{Fig_method/parpaings2.png} &
+%    \includegraphics[width=0.32\textwidth]{Fig_method/parpaings3.png}
+%  \end{tabular}
+%  \begin{picture}(1,1)(0,0)
+%    {\color{dwhite}{
+%      \put(-286,-25.5){\circle*{8}}
+%      \put(-171,-25.5){\circle*{8}}
+%      \put(-58,-25.5){\circle*{8}}
+%    }}
+%    \put(-288.5,-28){a}
+%    \put(-173.5,-28){b}
+%    \put(-60.5,-28){c}
+%  \end{picture}
+%  \caption{Automatic detection of blurred segments on a textured image.
+%           a) the input image,
+%           b) automatic detection result with initial assigned width set
+%           to 3 pixels,
+%           c) automatic detection result with initial assigned width set
+%           to 8 pixels.}
+%  \label{fig:noisy}
+%\end{figure}
 
 The automatic detection of blurred segments in a whole image is left
 available for testing in an online demonstration at the following address: \\

Code/Seg/BSTools/bsdetectionwidget.h

@@ -128,6 +128,11 @@ public:
    */
   void switchHighlightColors ();
 
+  /**
+   * \brief Switches the extraction stats display on or off.
+   */
+  inline void switchStats () { stats = ! stats; }
+
   /**
    * \brief Switches the extraction result display on or off.
    */
@@ -241,6 +246,8 @@ private:
   int background;
   /** Black level used to lighten background images. */
   int blevel;
+  /** Flag indicating whether detection stats should be output. */
+  bool stats;
   /** Flag indicating whether detection result should be output. */
   bool verbose;
 
@@ -348,6 +355,11 @@ private:
    */
   void displayBackground ();
 
+  /**
+   * \brief Writes the stats of the last detection in a file.
+   */
+  void writeStats ();
+
   /**
    * \brief Writes the result of the last detection in a file.
    */

Code/Seg/BlurredSegment/blurredsegmentproto.cpp

@@ -271,6 +271,7 @@ BlurredSegment *BlurredSegmentProto::endOfBirth ()
                                       DigitalStraightLine::DSL_THIN,
                                       xmin, ymin, xmax, ymax);
   }
+  else return (NULL);
   Pt2i aps (-1, -1), ape (-1, -1), apv (-1, -1);
   if (convexhull != NULL)
     convexhull->antipodalEdgeAndVertex (aps, ape, apv);

Code/Seg/DirectionalScanner/dynamicalscannero1.cpp

@@ -240,6 +240,12 @@ void DynamicalScannerO1::bindTo (int a, int b, int c)
     nu = (templ_nu * new_n1) / old_n1;
   else
     nu = (templ_nu * new_ninf) / old_ninf;
+  if (dlb < 0)  // dlb should stay positive to avoid the direction change
+  {             //   of the support line inequations.
+    dla = -dla;
+    dlb = -dlb;
+    c = -c;
+  }
   dlc1 = c + nu / 2;
   dlc2 = c - nu / 2;
 }