Usage
LiberTEM-blobfinder can evaluate the position of convergent beam electron diffraction disks or precession electron diffraction peaks to generate input data for strain maps in three automated steps:
Identify peaks using
get_peaks()Extract a probable lattice from the peak positions using
libertem_blobfinder.common.fullmatch.FullMatcher.full_match()Refine the lattice for each frame using
run_refine()
The algorithms are currently focused on the initial data extraction step, i.e. they work purely in pixels in frame coordinates and derive only parameters for each individual peak and for a 2D lattice in the detector reference frame. They don’t try to index peaks in crystallographic indices of the sample or to derive a 3D orientation. However, they can extract relevant input data such as peak positions and intensities very efficiently for such subsequent processing steps.
The algorithms are designed to be robust against intensity variations across a diffraction disk, between disks and between frames by calculating a correlation for each potential peak position, measuring the quality of each correlation and using the quality and position with subpixel refinement in a weighted least square optimization to derive the parameters for each frame. At the same time, the algorithms are optimized for efficiency and scalability so that they generate a full strain map on 32 GB of raw data within a few minutes on a suitable workstation and scale on a cluster, making full use of LiberTEM’s distributed processing capabilities.
Relevant input parameters are
- Matching template
Instance of
MatchPatternAvailable options are
RadialGradient,BackgroundSubtraction,RadialGradientBackgroundSubtraction, andUserTemplatesearchparameter to define the search area around the expected position
- Matcher: Instance of
libertem_blobfinder.common.gridmatching.Matcher. tolerancefor position errors in the matching routine
- Matcher: Instance of
Number of disks to find in the initial step
See Examples for practical application!