Changeset 16423 for trunk/psModules/src/imcombine/pmSubtraction.c

Timestamp:

Feb 13, 2008, 2:13:09 PM (18 years ago)

Author:

Paul Price

Message:

Starting to attempt to support stacking of images that have been read in piecemeal (scan by scan). Added parameters to pmSubtractionKernels to specify size of the (original) image, so that normalisation can be done (during convolution) when the image size is not the original image size. Doing some algebraic gymnastics in the pixel rejection to account for the image being smaller than the original.

File:

• trunk/psModules/src/imcombine/pmSubtraction.c (modified) (4 diffs)

trunk/psModules/src/imcombine/pmSubtraction.c

@@ -4 +4 @@
  *  @author GLG, MHPCC
  *
- *  @version $Revision: 1.77 $ $Name: not supported by cvs2svn $
- *  @date $Date: 2008-02-07 04:03:23 $
+ *  @version $Revision: 1.78 $ $Name: not supported by cvs2svn $
+ *  @date $Date: 2008-02-14 00:13:09 $
  *
  *  Copyright 2004-2007 Institute for Astronomy, University of Hawaii
@@ -715 +715 @@
 
     int numCols = ro1->image->numCols, numRows = ro1->image->numRows; // Image dimensions
+    int x0 = ro1->col0, y0 = ro1->row0; // Image offset
 
     psImage *convImage1 = out1->image;   // Convolved image
@@ -771 +772 @@
         yMax = PS_MIN(region->y1, yMax);
     }
+
+    // Size to use when calculating normalised coordinates (different from actual size when convolving
+    // subimage)
+    int xNormSize = (kernels->numCols > 0 ? kernels->numCols : numCols);
+    int yNormSize = (kernels->numRows > 0 ? kernels->numRows : numRows);
 
     psMaskType maskSource;              // Mask these pixels when convolving
@@ -798 +804 @@
     for (int j = yMin; j < yMax; j += fullSize) {
         int ySubMax = PS_MIN(j + fullSize, yMax); // Range for subregion of interest
-        float yNorm = 2.0 * (float)(j + size + 1 - numRows/2.0) / (float)numRows; // Normalised coordinate
+        float yNorm = 2.0 * (float)(j + y0 + size + 1 - yNormSize/2.0) /
+            (float)yNormSize; // Normalised coordinate
         for (int i = xMin; i < xMax; i += fullSize) {
             int xSubMax = PS_MIN(i + fullSize, xMax); // Range for subregion of interest
-            float xNorm = 2.0 * (float)(i + size + 1 - numCols/2.0) / (float)numCols; // Normalised coordinate
+            float xNorm = 2.0 * (float)(i + x0 + size + 1 - xNormSize/2.0) /
+                (float)xNormSize; // Normalised coordinate
 
             // Only generate polynomial values every kernel footprint, since we have already assumed