r326: FFTW additions, filter image generation

[ctsim.git] / libctsupport / mathfuncs.cpp

diff --git a/libctsupport/mathfuncs.cpp b/libctsupport/mathfuncs.cpp
index bc3b38308bfde55d9b6b2f7002d1801b15661b98..3604b2f5074ca360f5737882403038223a085c98 100644 (file)
--- a/libctsupport/mathfuncs.cpp
+++ b/libctsupport/mathfuncs.cpp
@@ -2,7 +2,7 @@
 **  This is part of the CTSim program
 **  Copyright (C) 1983-2000 Kevin Rosenberg
 **
-**  $Id: mathfuncs.cpp,v 1.1 2000/06/22 10:17:28 kevin Exp $
+**  $Id: mathfuncs.cpp,v 1.5 2001/01/01 10:14:34 kevin Exp $
 **
 **  This program is free software; you can redistribute it and/or modify
 **  it under the terms of the GNU General Public License (version 2) as
@@ -22,18 +22,18 @@
 
 
 /* NAME
- *    integrateSimpson         Integrate array of data by Simpson's rule
- *
- * SYNOPSIS
- *    double integrateSimpson (xmin, xmax, y, np)
- *    double xmin, xmax                Extent of integration
- *    double y[]               Function values to be integrated
- *    int np                   number of data points
- *                             (must be an odd number and at least 3)
- *
- * RETURNS
- *    integrand of function
- */
+*    integrateSimpson         Integrate array of data by Simpson's rule
+*
+* SYNOPSIS
+*    double integrateSimpson (xmin, xmax, y, np)
+*    double xmin, xmax         Extent of integration
+*    double y[]                Function values to be integrated
+*    int np                    number of data points
+*                              (must be an odd number and at least 3)
+*
+* RETURNS
+*    integrand of function
+*/
 
 double 
 integrateSimpson (const double xmin, const double xmax, const double *y, const int np) 
@@ -42,16 +42,16 @@ integrateSimpson (const double xmin, const double xmax, const double *y, const i
     return (0.);
   else if (np == 2)
     return ((xmax - xmin) * (y[0] + y[1]) / 2);
-
+  
   double area = 0;
   int nDiv = (np - 1) / 2;  // number of divisions
   double width = (xmax - xmin) / (double) (np - 1);    // width of cells
-
+  
   for (int i = 1; i <= nDiv; i++) {
     int xr = 2 * i;
     int xl = xr - 2;       // 2 * (i - 1) == 2 * i - 2 == xr - 2
     int xm = xr - 1;       // (xl+xr)/2 == (xr+xr-2)/2 == (2*xr-2)/2 = xr-1
-
+    
     area += (width / 3.0) * (y[xl] + 4.0 * y[xm] + y[xr]);
   }
   
@@ -63,13 +63,13 @@ integrateSimpson (const double xmin, const double xmax, const double *y, const i
 
 
 /* NAME
- *    norm_angle       Normalize angle to 0 to 2 * PI range
- *
- * SYNOPSIS
- *    t = norm_angle (theta)
- *    double t        Normalized angle
- *    double theta     Input angle
- */
+*    normalizeAngle       Normalize angle to 0 to 2 * PI range
+*
+* SYNOPSIS
+*    t = normalizeAngle (theta)
+*    double t         Normalized angle
+*    double theta     Input angle
+*/
 
 double 
 normalizeAngle (double theta)
@@ -81,3 +81,60 @@ normalizeAngle (double theta)
   
   return (theta);
 }
+\r
+\r
+void \r
+vectorNumericStatistics (std::vector<double> vec, const int nPoints, double& min, double& max, double& mean, double& mode, double& median, double& stddev)\r
+{\r
+  if (nPoints <= 0)\r
+    return;\r
+  \r
+  mean = 0;\r
+  min = vec[0];\r
+  max = vec[0];\r
+  int i;\r
+  int iMinPos = 0;\r
+  for (i = 0; i < nPoints; i++) {\r
+    double v = vec[i];\r
+    if (v > max)\r
+      max = v;\r
+    if (v < min) {\r
+      min = v;\r
+      iMinPos = i;\r
+    }\r
+    mean += v;\r
+  }\r
+  mean /= nPoints;\r
+  \r
+  static const int nbin = 1024;\r
+  int hist[ nbin ] = {0};\r
+  double spread = max - min;\r
+  mode = 0;\r
+  stddev = 0;\r
+  for (i = 0; i < nPoints; i++) {\r
+    double v = vec[i];\r
+    int b = static_cast<int>((((v - min) / spread) * (nbin - 1)) + 0.5);\r
+    hist[b]++;\r
+    double diff = (v - mean);\r
+    stddev += diff * diff;\r
+  }\r
+  stddev = sqrt (stddev / nPoints);\r
+  \r
+  int max_binindex = 0;\r
+  int max_bin = -1;\r
+  for (int ibin = 0; ibin < nbin; ibin++) {\r
+    if (hist[ibin] > max_bin) {\r
+      max_bin = hist[ibin];\r
+      max_binindex = ibin;\r
+    }\r
+  }\r
+  \r
+  mode = (max_binindex * spread / (nbin - 1)) + min;\r
+  \r
+  std::sort(vec.begin(), vec.end());\r
+  \r
+  if (nPoints % 2)  // Odd\r
+    median = vec[((nPoints - 1) / 2)];\r
+  else        // Even\r
+    median = (vec[ (nPoints / 2) - 1 ] + vec[ nPoints / 2 ]) / 2;\r
+}\r