Introduction
In this note we describe the PyCSPadImage python
-package which is intended to convert the CSPad raw data from HDF5 file to the geometry-corrected 2-D image array. Alternatively, this package evaluates and provide access to the pixel coordinate arrays. This package can be used in a stand-alone python
code.
Please note, that LCLS main-stream data analysis framework is based on C++ package Psana. To work with CSPad images Psana has several modules which are collected in the CSPadPixCoords package. Two of these modules, CSPadImageProducer and CSPadInterpolImageProducer, are intended to produce the 2-D array with CSPad image from raw data (currently XTC, but HDF5 will be available soon). Precise CSPad geometry is accounted in Psana using the calibration parameters supplied by the PSCalib package.
Package PyCSPadImage
Content
The python
-based PyCSPadImage package consists of modules (in alphabetic order):
CSPadConfigPars.py
- provides access to the CSPad configuration parametersCSPadImageProducer.py
- receives raw CSPad data 3-D (32, 185, 388) array, uses calibration parameters, produces the 2-D image arrayCalibPars.py
- provides access to the CSPad calibration parametersCalibParsDefault.py
- defines the default CSPad calibration parametersCalibParsEvaluated.py
- defines the evaluated CSPad parameters such as pixel coordinatesExamples.py
- contains examples of how to get the CSPad 2-D image array with corrected geometryAlignment.py
- similar to theExamples.py
, but is used for alignment purpose, its content is not guaranteedGlobalGraphics.py
- a set of useful global methods for interactive graphicGlobalMethods.py
- a set of miscellaneous global methodsHDF5Methods.py
- a set of global methods to work with HDF5 files
Functionality
The PyCSPadImage package is intended to convert the CSPad raw data from HDF5 file to the geometry-corrected 2-D image array. Methods of this package provide basic functionality as follows.
- Raw data 3-D (32, 185, 388) array can be be obtained from HDF5 file by its record name using methods from the module
HDF5Methods.py
. The CSPad dynamic configuration parameters are also uploaded using methods from the classCSPadConfigPars
.
- Access to the CSPad geometry calibration parameters are provided by the methods of the classes
CalibParsDefault
,CalibPars
, andCalibParsEvaluated
. It is assumed that the geometry calibration files, are available (see CSPad alignment) and located in the user-specified directory, otherwise default parameters will be used, which would not represent correctly the specific detector geometry.
- Methods of the class
CalibParsEvaluated
also evaluate pixel coordinates and provide access to the coordinate arrays of different shapes.
- Methods of the class
CSPadImageProducer
allows to get the geometry-corrected CSPad image or its part for specified 2x1 or quad.
How to get this package
Below we assume that all standard environment variable settings are done (otherwise see Analysis Workbook. Account Setup). In order to copy the PyCSPadImage package from SVN
repository and run a simple test one has to use commands:
log in to psana<XXXX> kinit cd <your-favorite-directory> newrel ana-current <your-release-directory-name> cd <your-release-directory-name> sit_setup addpkg PyCSPadImage HEAD cd PyCSPadImage/src/ <==== All source-code files are located here python Examples.py
External parameters
CSPAD geometry is varying for different detectors, experiments, or even runs. In order to keep track on all these variations LCLS offline has a simple calibration data base, which works as explained in CSPad alignment. In order to get correct CSPAD alignment parameters the pass to the calibration directory should be specified like this:
path_calib = '/reg/d/psdm/CXI/cxi80410/calib/CsPad::CalibV1/CxiDs1.0:Cspad.0'
If the detector configuration was changed during the experiment, then more than one calibration file should be available for the run ranges with stable configuration.
In order to access correct calibration file the run number should be provided, for example
runnum = 628
Data for CSPAD image and the detector configuration can be obtained from the HDF5 file, dataset name, and event number for example
fname = '/reg/d/psdm/CXI/cxi80410/hdf5/cxi80410-r0628.h5' dsname = '/Configure:0000/Run:0000/CalibCycle:0000/CsPad::ElementV2/CxiDs1.0:Cspad.0/data' event = 34
In further description we assume that this set of external parameters is defined.
Import modules
In code snippets below we use definitions of modules and libraries as follows
import numpy as np import CalibPars as calp import CalibParsEvaluated as cpe import CSPadConfigPars as ccp import CSPadImageProducer as cip import GlobalGraphics as gg import HDF5Methods as hm
Reconstruction of CSPAD image
Entire code example for image reconstruction is
calp.calibpars.setCalibParsForPath ( run=runnum, path=path_calib ) ds1ev = hm.getOneCSPadEventForTest( fname, dsname, event ) cspadimg = cip.CSPadImageProducer(rotation=0, tiltIsOn=True, mirror=True) arr = cspadimg.getCSPadImage( ds1ev )
First, one has to set the correct version of the calibration parameters
calp.calibpars.setCalibParsForPath ( run=runnum, path=path_calib )
Then, one need in CSPAD dataset for event,
ds1ev = hm.getOneCSPadEventForTest( fname, dsname, event )
this method returns the CSPAD data as a numpy array for one event, ds1ev.shape=(Nseg, 185, 388), where Nseg?32.
Then, one has to initialize the object of the class CSPadImageProducer
cspadimg = cip.CSPadImageProducer(rotation=0, tiltIsOn=True, mirror=True)
with optional parameters
rotation
integer from 0 to 3 parameters for CSPAD orientation as 90*rotation
degree.tiltIsOn
=True
orFalse
- to account or not the tiny tilt angle of 2x1 sections.mirror
=True
orFalse
- to mirror reflect or not the image.
Finally the method
arr = cspadimg.getCSPadImage( ds1ev )
returns the 2-d numpy array with CSPAD image, which can be plotted using for example matplotlib
.
CSPAD pixel coordinate arrays
CSPAD pixel coordinate arrays can be evaluated/returned in two different shapes:
- for entire CSPAD with shape=(4,8,185,388)
- for data-driven shape=(Nseg,185,388), where Nseg?32 if some quads/segments are missing in data.
To get pixel coordinate arrays shaped for entire CSPAD use code:
calp.calibpars.setCalibParsForPath ( run=runnum, path=path_calib ) cpe.cpeval.evaluateCSPadPixCoordinates (rotation=0) xpix, ypix = cpe.cpeval.getCSPadPixCoordinates_pix()
where xpix and ypix are the coordinate (in pixels) arrays with shape = (4,8,185,388).
Method xpix_um, ypix_um = cpe.cpeval.getCSPadPixCoordinates_um()
returns pixel coordinates in micrometer.
To get CSPAD pixel coordinate arrays shaped as in data use code:
calp.calibpars.setCalibParsForPath ( run=runnum, path=path_calib ) cpe.cpeval.evaluateCSPadPixCoordinatesShapedAsData(fname,dsname,rotation=0) xpix, ypix = cpe.cpeval.getCSPadPixCoordinatesShapedAsData_pix()
where xpix and ypix are the coordinate (in pixels) arrays with shape = (Nseg,185,388).
Note that the fname
and dsname
need to be specified in order to get configuration of the data array.
Method xpix_um, ypix_um = cpe.cpeval.getCSPadPixCoordinatesShapedAsData_um()
returns pixel coordinates in micrometer.
The coordinate arrays extracted for both shapes are tested in module Examples.py
by the methods
example_of_image_built_from_pix_coordinate_array_shaped_as_data()
and
example_of_image_built_from_pix_coordinate_array_for_entire_cspad()
,
where images are reconstructed through the pixel coordinate arrays in
cpe.cpeval.getTestImageShapedAsData(ds1ev)
and
cpe.cpeval.getTestImageForEntireArray(ds1ev)
, respectively.
The last two methods use implicit loops over all pixels, that works pretty slow in Python. These modules are used for test only and are not recommended for real applications.
Example
Module Examples.py
demonstrates how to use the PyCSPadImage package. The essential part of the example can be presented as:
import sys import os import CalibPars as calp import CSPadConfigPars as ccp import CSPadImageProducer as cip import GlobalGraphics as gg # For test purpose in main only import HDF5Methods as hm # For test purpose in main only #---------------------------------------------- def main_example_xpp() : print 'Start test in main_example_xpp()' path_calib = '/reg/d/psdm/xpp/xpp47712/calib/CsPad::CalibV1/XppGon.0:Cspad.0' fname, runnum = '/reg/d/psdm/xpp/xpp47712/hdf5/xpp47712-r0043.h5', 43 dsname = '/Configure:0000/Run:0000/CalibCycle:0000/CsPad::ElementV2/XppGon.0:Cspad.0/data' event = 0 print 'Load calibration parameters from', path_calib calp.calibpars.setCalibParsForPath ( run=runnum, path=path_calib ) print 'Get raw CSPad event %d from file %s \ndataset %s' % (event, fname, dsname) ds1ev = hm.getOneCSPadEventForTest( fname, dsname, event ) print 'ds1ev.shape = ',ds1ev.shape print 'Make the CSPad image from raw array' cspadimg = cip.CSPadImageProducer(rotation=0, tiltIsOn=True, mirror=False) arr = cspadimg.getCSPadImage( ds1ev ) print 'Plot CSPad image' gg.plotImage(arr,range=(0,2000),figsize=(11.6,10)) gg.move(200,100) gg.plotSpectrum(arr,range=(0,2000)) gg.move(50,50) print 'To EXIT the test click on "x" in the top-right corner of each plot window.' gg.show()
Let us consider in detail what needs to be done in order to produce the CSPad image.
First, all necessary modules need to be imported:
import CalibPars as calp import CSPadConfigPars as ccp import CSPadImageProducer as cip import GlobalGraphics as gg # For test purpose in main only import HDF5Methods as hm # For test purpose in main only
Then, the path to the calibration data types, the HDF5 data file name, and the dataset name in HDF5 structure need to be defined:
path_calib = '/reg/d/psdm/xpp/xpp47712/calib/CsPad::CalibV1/XppGon.0:Cspad.0' fname = '/reg/d/psdm/xpp/xpp47712/hdf5/xpp47712-r0043.h5' dsname = '/Configure:0000/Run:0000/CalibCycle:0000/CsPad::ElementV2/XppGon.0:Cspad.0/data'
- In principal, the path to the calibration data types can be defined automatically, but there is a chance that user would want to keep calibration files in non-standard place.
- The CSPad dataset can be also found automatically, but there might be more than one CSPad detector in experiment...
Then, calibration parameters need to be loaded:
calp.calibpars.setCalibParsForPath ( run = 1, path = path_calib )
- Currently, the calibration parameters are defined through the "singleton" object
calp.calibpars
.
If more than one CSPad is going to be used simultaneously, then different sets of calibration parameters
need to be loaded and thecalp.calibpars
needs to be de-referenced whenever it is necessary. - Run number needs to be specified if there are more than one calibration file available for different run ranges.
Then, the raw CSPad dataset needs to be extracted:
ds1ev = hm.getOneCSPadEventForTest( fname, dsname, event )
It is recommended to use this method, which also loads correct configuration parameters from HDF5 file.
Default version of the configuration parameters can be wrong, if not all 2x1 are in use.
Finally the 2-D image array can be obtained as a numpy array
cspadimg = cip.CSPadImageProducer(rotation=0, tiltIsOn=True, mirror=False) arr = cspadimg.getCSPadImage( ds1ev )
and used, for example for plotting
gg.plotImage(arr,range=(0,2000),figsize=(11.6,10)) gg.plotSpectrum(arr,range=(0,2000)) gg.show()