DOCUMENTATION CIV PROCESSING :
|
Open the web page, presently http://z4/civ To display the current queue file http://z4/civ/civbot/TODO.txt ; right-click to refresh For Patch-UP processing goto |
FAQ : Q : In the “Processing option” page I have the “cannot open file” message . A : Check if you use a network path as shown on figure 1 Check the permission on your files ( and contact your admin … ) Q : How can I have access to the program output ? A
: By selecting the “OUTPUT |
On this first page you can select which experiment you would like to process. An experiment is defined by its configuration file : the .CIV file http://www.civproject.org/DOC%5Cfile_format_CIV.htm
If you are using a distributed version you HAVE TO use network path (UNC) for your selected files (as: \\computer-name\shared-folder).
In order to reduce network and drive space usage, please use
compressed image in place of non-defined raw images filled with 25% of 0 ! (HOWTO
You may also select a grid file and/or a mask image file. The grid file lets you select “exactly” ( please read the convect flow discussion for further understandings ) the physical location of each node. The mask file lets you define area in the image where it would be useless to process displacement.
For grid specifications : http://www.civproject.org/DOC%5Cfile_format_GRID.htm
For mask specifications : http://www.civproject.org/DOC%5Cfile_format_mask.htm
After appropriate files selection and a click on the processing options button you access to this new html page :
The first lines :
confirms the successful access to the .civ file. If it cannot access to the civ file, it will be noticed here and the Image organization buttons won’t be initialized with numbers.
The Algorithm
selection box :
Here you select which algorithm you want to run ; CIV1 is for the first run, robust code and it needn’t guess displacement field.
If you want to use advanced techniques as peak-locking minimization or image deformations you have to select CIV2.
The beta flavor are the most advanced and optimized codes ……
Algorithm options :
in this box you may select different processing options which drives the processing.
HART : available with CIV1, informations on this page : http://www.civproject.org/DOC/hart.htm
This option permits to reduce the number of falses vectors and has to be used for first pass.
DECIMAL SHIFT : available with CIV2
As described in the literature, peak-locking effect is minimum for integer displacement measurement. According to the guess displacement field we translate the search box on the second image ( which is in fact a continuous function after interpolation) by the decimal part of the guess displacement . Hence if the guess field was accurate “enough” this would lead to integer displacement measurement with reduce PL errors component.
A more proper implementation of this technique would be to use bilinear interpolation for images instead of the heavy TPS scheme.
Also instead of translating only the second image it’s better to translate each of them by the half translation and correct the coordinates of the estimation location.
DEFORMATION : available with CIV2
Blah Blah Blah
REVERSE CORRELATION :
Not yet
POWER 4 :
Not yet
HOLE
:
If a vector is ‘suspect ‘ he will not be written in the .v file.
Boxes size parameters
In this sub-section, you select the appropriate correlation and search region size .
IBx & IBy are the dimension in pixels in the x & y directions of the correlation box.
ISx & ISy are the correspondings for the search area.
Ro is the smooth coefficient used in the correlation interpolation scheme to smooth the function.
If you didn’t choose an explicit mask file , the algorithm will build a regular grid with the grid spacing defined .
Image organisation
You can explicitly choose which images you want to process, the present definition is :
We process the “First Image”, “First Image+N Step”…. Until the “Last Image’
Volume size is not yet implemented , if you need special arrangement just let me know
Global shift
If the flow you’’re stunding is convected by a mean flow, you can substract this mean flow and limit the search area. For example in a turbulent channel with a mean flow corresponding to 30 pixels in y direction, you can specifie this value here and improve the processing speed and accuracy.
Use an existing .v file : if you have a previous knowledge of the velocity field and wrote guess velocity files you can use them by checking this option.
Image selection
As we suggest for the acquisition, if you took severals images in each burst you are able to select which images pairs you want to process.
The delta time between the images is given for the first burst.
FIX
Description of the algorithm : FIX on civproject.org
To select the False Vector
Detector during CIV processing ,
you have to check the yes box.
The signification
of the differents parameters is explained
on the Algorithm
page
You can also run
a Fix on previous CIV results by checking the FIX ONLY box.
In this case don’’t forget to check the correct
images pairs…………