osiv_corr

Langue: en

Version: 149883 (fedora - 04/07/09)

Section: 1 (Commandes utilisateur)

NAME

osiv_corr - perform particle image velocimetry

SYNOPSIS

osiv_corr [ -o output ] [ -p file] movie_a [movie_b]

DESCRIPTION

osiv_corr is a program for performing efficient analysis of particle image velocimetry (PIV) images. It provides a variety of algorithms for pre-processing, interpolation, and cross correlation of images. This description is intended as a quick reference, more information can be found in the online manual at (www.osiv.org).

Parameters can be passed to osiv_corr using two methods. In the first, parameters are passed using options on the command line. In the second, parameters are placed in a text file and the filename is passed using the -p option. These parameters control the methods used to perform the correlation and the nature of the output.

The output of osiv_corr is a binary file that contains the displacement vectors of particles. These displacement vectors can be visualized using osiv_draw or printed in text form using osiv_dump.

PARAMETER FILE

The parameter file is specified using the -p option. This is a text file which contains parameters in addition to those which are specified on the command line. The format of this file is simple. Each line specifies a single parameter, for example:


    xcorr_alg = 5
    grid_xsize = 32
    grid_ysize = 32
    ...

The parameter name used in this file is very similar to that used on the command line (see PARAMETERS). The only difference being that `-' characters in the command line name are replaced with `_' characters in the parameter file. Additional white space is allowed. Blank lines and lines beginning with `#' are ignored.

Parameters from a parameter file are "weak" and may be overwritten by parameters that occur on the command line or that occur in later parameter files. Parsing of parameters occurs in the order which the arguments occur in the command line.

MOVIE FILES

The images which are correlated in order to calculated displacement vectors are taken from movies specified on the command line. If one filename is present, both the first and second frames are taken from the same movie. If two filenames are present, then the first frame is taken from the first movie and the second frame is taken from the second movie. The movie type is inferred from the file extension, possible file extensions are .dat, .tif, .tiff. The manual contains additional information.

OUTPUT FILE

The output file is specified with the -o option. If this option is omitted, data will be written to the the file "osiv.out". The output file contains the calculated displacement vectors as well as all of the parameters which were passed to osiv_corr. This file has a binary format, but can be interpretted using osiv_dump.

PARAMETERS

-l --list
Reasonable default values are automatically chosen for all parameters based on the dimensions of the image and those parameters which are explicitly specified. This option causes osiv_corr to print out the chosen parameters and then exit without performing the correlation.
--xcorr-alg INTEGER
The algorithm used to perform the correlation is selected by specifying one of the following integers:
 1 - Direct Least Squares
 2 - Fast Direct Least Squares
 3 - Direct Correlation
 4 - Fast Direct Correlation
 5 - Fourier Correlation
 6 - Fourier Least Squares
 7 - Iterative Fourier Correlation 
--pproc-alg INTEGER
The algorithm used to perform pre-processing of the images is selected by specifying one of the following integers:
 0 - None
 1 - Subtract to minima
 2 - Add to maxima
 3 - Stretch to limits
 4 - Stretch to variance
 5 - Stretch to average
--interp-alg INTEGER
The algorithm used to perfom peak interpolation is selected by specifying one of the following integers:
 0 - None
 1 - Gaussian
 2 - Parabolic
 3 - Paraboloidal
 4 - Centroid
--image-xsize INTEGER
The width of the images must be the same for every frame. As osiv_corr can almost always determine image sizes from the input image files, this option is very rarely required. Upper bounds on the image size are specified during the compilation.
--image-ysize INTEGER
The height of the image must be the same for every frame. As osiv_corr can almost always determine image sizes from the input image files, this option is very rarely required. Upper bounds on the image size are specified during the compilation.
--grid-xstart INTEGER
The horizontal position of the first correlation window on the image frame. A value of zero abuts the left edge of the first window against the left edge of the frame. You should note that certain algorithms must be able to access the pixels surrounding the window, so the first window may need to be offset. Please see the manual for more information.
--grid-ystart INTEGER
The vertical position of the first correlation window on the image frame. A value of zero abuts the top edge of the first window against the top edge of the frame. You should note that certain algorithms must be able to access the pixels surrounding the window, so the first window may need to be offset. Please see the manual for more information.
--grid-xspace INTEGER
The horizontal spacing between correlation windows on the image frame. A value of one places overlapping windows on every pixel.
--grid-yspace INTEGER
The vertical spacing between correlation windows on the image frame. A value of one places overlapping windows on every pixel.
--grid-xsize INTEGER
The number of windows placed in each row of the grid of correlations windows used for each frame. The grid size must not extend the grid beyond the image bounds, see the manual for more information.
--grid-ysize INTEGER
The number of windows placed in each column of the grid of correlations windows used for each frame. The grid size must not extend the grid beyond the image bounds, see the manual for more information.
--disp-xoffset INTEGER
A horizontal displacment around which all correlations are calculated. Adding a fixed displacement improves the accuracy of the correlation in flow fields that are primarily uniform. The displacement is added back to the vector before values are output, so no post-processing is required.
--disp-yoffset INTEGER
A vertical displacment around which all correlations are calculated. Adding a fixed displacement improves the accuracy of the correlation in flow fields that are primarily uniform. The displacement is added back to the vector before values are output, so no post-processing is required.
--disp-xmax INTEGER
The maximum horizontal displacement to be examined. For direct methods, the correlation is only evaluated up to this limit. For FFT-based methods, the search is limited to this maximum.
--disp-ymax INTEGER
The maximum vertical displacement to be examined. For direct methods, the correlation is only evaluated up to this limit. For FFT-based methods, the search is limited to this maximum.
--disp-xiter INTEGER
The maximum horizontal displacement to which the Iterative Fourier Correlation algorithm may displace a window. It is ignored by all other algorithms. Note that the correlation window will then be limited to a maxmimum displacement of disp_xmax + disp_xiter.
--disp-yiter INTEGER
The maximum vertical displacement to which the Iterative Fourier Correlation algorithm may displace a window. It is ignored by all other algorithms. Note that the correlation window will then be limited to a maxmimum displacement of disp_ymax + disp_yiter.
--wind-xsize INTEGER
Horizontal size of the window used to perform the correlation. Larger windows are generally less prone to error, while smaller windows yield finer spatial resolution. For FFT-based methods, powers of two allow efficient algorithms.
--wind-ysize INTEGER
Vertical size of the window used to perform the correlation. Larger windows are generally less prone to error, while smaller windows yield finer spatial resolution. For FFT-based methods, powers of two allow efficient algorithms.
--wind-size INTEGER
Set the horizontal and vertical sizes of the window to the same value, simultaneously. This is identical to calling both --wind-xsize and --wind-ysize with the same value.
--wind-xsuper INTEGER
Horizontal size of the larger underlying correlation window used in the Fourier Correlation and Fourier Least Squares algorithms. This parameter is ignored by all other algorithms. See the manual for more information.
--wind-ysuper INTEGER
Vertical size of the larger underlying correlation window used in the Fourier Correlation and Fourier Least Squares algorithms. This parameter is ignored by all other algorithms. See the manual for more information.
--wind-super INTEGER
Set the horizontal and vertical size of the super window to the same value, simultaneously. This is identical to calling both --wind-xsuper and --wind-ysuper with the same value.
--movie-start INTEGER
The first frame of the movie on which cross correlation is performed. This number is used to calculate the filename to use with some movie types, the tiff directory to read with tiff movies, and byte offsets for raw movies. See the manual for more information.
--movie-finish INTEGER
The last frame of the movie on which cross correlation is performed. This number is used to calculate the filename to use with some movie types, the tiff directory to read with tiff movies, and byte offsets for raw movies. See the manual for more information.
--movie-skip INTEGER
The number of frames to skip between sets of frames that are cross correlated. See the manual for more information and some helpful examples.
--movie-set INTEGER
The number of frames that separate a pair of images to be cross correlated. See the manual for more information and some helpful examples.
--num-iter INTEGER
The number of iterations to perform when using the iterative cross correlation algorithm. All other algorithms ignore this parameter.
--fft-wisdom FILE
The FFT-based algorithms make use of the FFTW library. This library is self-tuning and can make use of timing information which may be generated by the program osiv_tune.
--write-map
This option causes correlation maps to be written to be included in the output file. This substantially increases the size of the output file, and so this output should normally be omitted.

MORE INFO

Additional documentation can be found online at www.osiv.org.

SEE ALSO

osiv_dump(1), osiv_draw(1), osiv_tune(1), osiv_synth(1),
readovd(3).

BUGS

No known bugs. Please report bugs as www.osiv.org. Copyright © 2003-2005 James Strother.