piv_handling#

Example Usage#

Full Contents:#

piv_handling.GetPIV(im1, im2, X1, Y1, EdgeLength, histequilize)#

GetPIV computes PIV flow field estimate based on image1 (im1) and image2 (im2) using the Phase Correlation method implemented in xcorr2fft. im1 & im2 are assumed to have the same dimensions. The grid X1,Y1 is assmued to be contained in the image domain with finite EdgeLength defining size of PIV box. Output are components of the flow fiel on the grid

Written by: Sebastian J Streichan, KITP, February 01, 2013 NPM added histeq step (optional)

piv_handling.PIVScript#

Using GetPIV we computes the flow field estimate based on an image sequence im1 & im2 are assumed to have the same dimensions. The grid X1,Y1 is assmued to be contained in the image domain with finite EdgeLength defining size of PIV box. This script generates a movie displaying the original image with flow field overlayed.

Written by: Sebastian J Streichan, KITP, February 14, 2013 NPM added histequalize option on 27

piv_handling.PIVTimeseries(inputDir, options)#

Using GetPIV we computes the flow field estimate based on an image sequence im1 & im2 are assumed to have the same dimensions. The grid X1,Y1 is assmued to be contained in the image domain with finite EdgeLength defining size of PIV box. This script generates a movie displaying the original image with flow field overlayed.

Written by: Sebastian J Streichan, KITP, February 14, 2013 NPM added histequalize option and made function

piv_handling.advectRuntStripes#

test script

piv_handling.advectRuntStripes_script#

test script

piv_handling.applyOpticalFlow(pivStack, im0, Options)#
Parameters

  • Options (struct with fields) –

    mandatory fields:

    outDir minT maxT

    optional fields:
    plot_texturepathsbool (default=true)

    make movie of optical flow

    plot_scatterpathsbool (default=false)

    make movie of tracer beads

    clipImagebool (default=true)

    clip the texturepatch image to a rectangle

    szX : size of resampled images in which PIV is performed szY : size of resampled images in which PIV is performed X0 : evaluation grid points, x coord Y0 : evaluation grid points, y coord

  • 2021 (NPMitchell) –

piv_handling.extractMetricFromPIVPair(VX, VY, WX, WY, computation_method, truncation_factor)#

Extract a comparison metric from computation on two 2D PIV fields V,W returning a matrix of metric values calculated at all points in the field. VX, VY, WX, WY are assumed to have the same dimensions. The field will be spatially truncated if truncation_factor is not 0. The method of computation used is specified by piv_computation_method. The dimension to truncate is given by truncation_dim (1 or 2) The factor to truncate the image by, on both sides, is given by truncation_factor. The default dimension is 2, the horizontal axis of the image i.e. the vertical axis of the PIV field (the PIV program transposes the image when computing PIV to be symmetric about the vertical axis). Also returns the minimum and maximum time indices used along the truncation dimension, the matrix values outside of that range should all be zeros.

Modified on 8/31 to input V_RMS and W_RMS as the RMS values for the given fields, where smoothed and truncated RMS values can be taken into account rather than simply the standard calculation.

Parameters

  • VX (X component of first PIV field (V), 2D matrix) –

  • VY (Y component of first PIV field (V), 2D matrix) –

  • WX (X component of second PIV field (W), 2D matrix) –

  • WY (Y component of second PIV field (W), 2D matrix) –

  • computation_method (string with potential values) – normalized residual unnormalized residual inner product

  • truncation_dim_img (dimension to truncate over, int with value 1 or 2,) –

  • 2 (default dimension) –

  • truncation_factor (double factor to truncate matrix by on both sides,) –

  • (exclusive) (must be between 0 (inclusive) and 0.5) –

  • Returns – MetricMat, a 2D matrix of correlation values calculated based on the chosen method.

  • Outputs –

  • ------- –

  • 2020 (Vishank Jain-Sharma) –

piv_handling.getPIVLab(im1, im2, options)#

[uu, vv, u_filt, v_filt, xx, yy] = getPIVLab(im1,im2,options)

getPIVLab computes PIV flow field estimate based on image1 (im1) and image2 (im2) using the Phase Correlation method implemented in PIVLab. im1 & im2 are assumed to have the same dimensions. Output are components of the flow field on the grid (xx, yy)

Parameters

  • im1 (NxM numeric) – image at t0

  • im2 (NxM numeric) – image at t0 + deltaT

  • options (optional struct with optional fields) – intArea1 subpixFindr mask roi numPasses intArea2 intArea3 intArea4 repeat disAuto clahe claheW highPass highPassSz clipping wiener wienerW valid_vel do_stdev_check stdthres do_local_median neigh_thres

Returns

  • uu (PxQ float) – velocity in x

  • vv (PxQ float) – velocity in y

  • u_filt (PxQ float) – velocity in x

  • v_filt (PxQ float) – velocity in y

  • xx (PxQ float) – evaluation points in x

  • yy (PxQ float) – evaluation points in y

    Written by NPMitchell 2022

piv_handling.processMedFiltersScript#

script to run med filters on PIV data

piv_handling.pullbackPathlines(QS, x0, y0, t0, options)#

Create paths in pullback space (in pixels, XY) by following optical flow of PIV measured on standardized pullback map. Note: I’ve chosen to use spatial smoothing via Gaussian blur rather than temporal smoothing of velocities for this code.

Parameters

  • pivStack (struct with two #TP x nX x nY float entries) –

    vx: #TP x nX x nY float

    velocities in x direction

    vy: #TP x nX x nY float

    velocities in y direction

  • x0 (n*m float array) – x coordinates in pullback pixels to start pathlines at t0

  • y0 (n*m float array) – y coordinates in pullback pixels to start pathlines at t0

  • t0 – time at which to begin the pathlines, must be member of QS.xp.fileMeta.timePoints

  • options (struct with fields) –

    previewbool

    view intermediate results

    timePoints1d int array

    the timpepoints along which to map pathlines that intersect with (x0, y0) at t0

Returns

  • XX (#timePoints x size(x0, 1) x size(x0, 2) float array) – the pullback pixel X coordinates of the pathline spanning timePoints

  • YY (#timePoints x size(x0, 1) x size(x0, 2) float array) – the pullback pixel Y coordinates of the pathline spanning timePoints

  • NPMitchell 2021

piv_handling.rk4DynamicVelocity2D(pts, X0v, Y0v, vx, vy, hh, tt)#

[xyOut, xyPaths] = rk4DynamicVelocity2D(pts, XY0v, v2d, hh, tt) solve Runge-Kutta 4th order in 2d space with fixed velocity field

Parameters

  • pts (Px2 numeric) – points at which we begin tracing paths in 2d

  • XY0v (MxN numeric) – evaluation points at which velocities reside

  • vx (TxMxN numeric) – dynamic 2d velocity in x at evaluation points (time, vxgridX, vxgridY)

  • vy (TxMxN numeric) – dynamic 2d velocity in y at evaluation points (time, vygridX, vygridY)

  • hh (optional float 0<h<1 (default=0.05)) – stepsize for finding path

  • tt (optional Qx1 float array (default=0:hh:size(v2d, 1))) – timestamps relative to the indices of v2d at which to compute

Returns

  • xyOut (Px2 float) – final positions

  • xyPaths (length P cell array) – paths from pts to xyOut

  • NPMitchell 2022

piv_handling.rk4FixedVelocity2D(pts, X0v, Y0v, vx, vy, hh, tt)#

[xyOut, xyPaths] = rk4FixedVelocity2D(pts, XY0v, v2d, hh, tt) solve Runge-Kutta 4th order in 2d space with fixed velocity field

Parameters

  • pts (Px2 numeric) – points at which we begin tracing paths in 2d

  • XY0v (MxN numeric) – evaluation points at which velocities reside

  • v2d (MxN numeric) – 2d velocity at evaluation points

  • hh (float 0<h< 1) – stepsize for finding path

Returns

  • xyOut (Px2 float) – final positions

  • xyPaths (length P cell array, such that xyPaths{i} is #ttx2 float array) – paths from pts to xyOut

  • NPMitchell 2022

piv_handling.runMedianFilterOnPIVField(input_PIV_path, output_path)#

Runs a median filter on the PIV data located in input_PIV_path and saves the data to output_path

Parameters

  • input_PIV_dir (path to directory containing all the PIV .mat structures) –

  • output_dir (path to directory in which to save median filtered PIV) –

  • median_order (order of median filter--total range to take median over) –

  • Outputs –

  • fields (new directory with name output_directory that containts saved PIV) –

  • filter (which have been run through median) –

Vishank Jain-Sharma 2020, edits NPMitchell 2022

piv_handling.xcorr2fft(image1, image2)#

xcorr2fft computes offsets between images image1 and image2 based on Phase Correlation method. image1 & image2 are assumed to have the same dimensions.

Written by: Sebastian J Streichan, EMBL, February 29, 2012

Indices and tables#