Changeset 29001 for trunk/Ohana/src/relastro/doc/notes.txt

Timestamp:

Aug 20, 2010, 12:05:07 PM (16 years ago)

Author:

eugene

Message:

optimizations for dvomerge and relastro; complete checks of db I/O success; fix overloaded image flags

Location:

trunk/Ohana

Files:

• . (modified) (1 prop)
• src/relastro/doc/notes.txt (modified) (1 diff)

trunk/Ohana/src/relastro/doc/notes.txt

@@ +1 @@
+
+2010.07.15
+
+  Some issues:
+
+  * when I apply the fitted image parameters (UpdateObjectOffsets), I
+    currently check for good/bad fits and warn if the image is bad.
+    But the correction requires multiple re-loads of the catalog
+    tables.
+
+  * I would like to check for bad fits by defining a grid of test
+    points and seeing how much they move.  However, for some chips,
+    the region of valid data is much smaller than the regino with any
+    data.  
+
+  * I would like to define a grid (say, Nx * Ny with Nx >> polynomial
+    order) and only use those grid regions in testing the solution
+
+  * This means knowing which detections belong on a specific image.
+    I'm not sure if there is a good way to do this.  at the moment, it
+    is easy to do detection -> image, but not so easy to do
+    image->detection.
+
+  * That might argue for generating the grid for each image, then
+    running through all detections and, as detections are examined,
+    populate the grid elements for their own image.
 
 2010.03.24