Changeset 32689 for branches/eam_branches/ipp-20111110/psphot/src/psphotFitSourcesLinear.c

Timestamp:

Nov 17, 2011, 1:09:10 PM (15 years ago)

Author:

eugene

Message:

correctly handle the second pass in the detection image: the code was creating duplicate detections in 2 ways: 1) the source subtraction before the second pass was only being applied to the SRC, not the DET image, thus all 1st pass detections were also found as 2nd pass detections; 2) the DBL star fitting function was causing trouble, with the two positions merging to a single object -- this was considered a success, but left behind flux in the subtracted image (and garbage detections in general); Ive turned off the DBL fitting for now as I am not convinced that code is ready

File:

• branches/eam_branches/ipp-20111110/psphot/src/psphotFitSourcesLinear.c (modified) (1 diff)

branches/eam_branches/ipp-20111110/psphot/src/psphotFitSourcesLinear.c

@@ -342 +342 @@
     psLogMsg ("psphot.ensemble", PS_LOG_MINUTIA, "solve matrix: %f sec (%d elements)\n", psTimerMark ("psphot.linear"), sparse->Nelem);
 
-    // XXXX **** philosophical question:
-    // we measure bright objects in three passes: 1) linear fit; 2) non-linear fit; 3) linear fit:
-    // should retain the chisq and errors from the intermediate non-linear fit?
-    // the non-linear fit provides better values for the position errors, and for
-    // extended sources, the shape errors
+    // Philosophical question: we measure bright objects in three passes: 1) linear fit; 2)
+    // non-linear fit; 3) linear fit: should we retain the chisq and errors from the
+    // intermediate non-linear fit?  the non-linear fit provides better values for the position
+    // errors, and for extended sources, the shape errors
 
     // adjust I0 for fitSources and subtract