Edited: 1-Jul-2015
							   
         How to Analyze Blobs using Bill Davis software tools

(A similar narative, with sample images, can be found at 
http://nstx.pppl.gov/nstx/Software/Applications/BlobTracking.html)

Blob analysis can be done automatically by running programs (like loadblobs.pro)
to load an SQL database, or produce tab-delimited text files of many blobs. Each
row represents characteristics of an identified blob in a particular camera
frame. This blob identification is a two- or three-step technique where first 
each frame is analyzed for blobs, and then frames are compared (by tracking the
first occurrance of a blob in subsequent frames) and parent-child relationships
are added to the data table. Then, these parent-child relationships are used
to compute the blob velocities. 

When loading blobs, various criteria are used for determining whether to count 
a closed contour in a processed fast camera image as a blob. Values for these
criteria are determined subjectively by playing the movies in FCplayer.pro, 
enhancing the image in various ways (such as by smoothing, normalizing, 
changing settings under "Edit...Change Gamma, Min/Max, Saturation...", etc.) and
adjusting the images until satisfied.

Defaults used by Bill Davis in March, 2014 for NSTX Fast Camera normalized
images are:

   minht=1.2
   minRise = 0.06
   ChiSq = 5	(a somewhat arbitrary number indicating how well a contour
                 fits an ellipse. 8 was used sometimes.)
   maxJump = 10.
   minArea = 15.
   minLifetime = 3

For the reloading of the database of the 150 "top" shots from 2010, a minHt of
1.1 was used. When a table was made for Stewart Zweben to create figures,
additional constraints were imposed: minHt >= 1.5 and parent ne -1 (because
velocity is zero for these).

The blob database, for the 2010 shots at least, was created from "normalized"
images. This is where each frame is divided by an average frame. The averaging
time naturally affects the results, especially when there are different types
of things happening over the time in question. The loadblobs.pro code reads
the peaktime from GPI_2010_Top_Shots.txt, and sets the averaging time as +/- 
5 msec around the time of the peak of the GPI signal. The default in FCplayer,
on the other hand, doesn't know about this peak time and computes a different
average frame every millisecond. When using FCplayer to check database blob
characteristics, the AveT1 and AveT2 keywords should be set to the peaktime
+/- 5 ms for similar results.

The SQL tables in the Blob database can be examined, constrained, used in plots,
etc., by:

   IDL> dbaccess, 'blobs'

The shot numbers are in the table names. All the columns in the Blob SQL tables
are described in  http://w3.pppl.gov/~bdavis/swdoc/blobDBdefs.txt.

To create these tables, or files containing the data, you can use loadblobs.pro.
It calls saveblobs.pro (which can write a file with the blob information 
computed for individual frames) and then populateblobs.pro. Using keyword 
outFileName when calling loadblobs.pro (or  populateblobs.pro) gives you a file
of what would go in the database. 

For example, 

   IDL> loadblobs, 139432, T1=0.270, T2=0.280, outDBfile='blobs_139432.txt'
   
will fill blobs_139432.txt with all the blob information, including speeds, 
parent/child relationships, etc. /u/bdavis/GPI/call_loaddb.pro can be used
to load all the shots in 
/w/nstx.pppl.gov/htdocs/nstx/Software/GPI/GPI_2010_Top_Shots.txt, or a range 
within it.

populateblobs.pro reads a file of blob characteristics created from fcplayer.pro
or saveblobs.pro, which is usually called, together with populateblobs, from
loadblobs.pro. I.e.,

--loadblobs
  --saveblobs
    --findblobs
      --fcfitellipse (returns tilts in radians, clockwise from zero to left)
        --mpfitellipse
      --blobfilewrite (writes file with tilts in radians, CW zero to left)
  --populateblobs (reads file, and converts tilts to degrees)

The graphs composing the "GPI Data Summary" pages can be created by programs
in /u/bdavis/GPI.

See http://nstx.pppl.gov/nstx/Software/Applications/BlobSpeedValidation.html 
for details and examples on how blob speed calculations
    were checked and verified.

Also see /u/bdavis/w3_html/swdoc/Blob_Tilt_Calcs.txt for ways to validate
blob tilt calcuations, and what conventions are used.