Page History

...

Image peak finding

Here are a collection of useful algorithms for image analysis: http://docs.scipy.org/doc/scipy/reference/ndimage.html

...

• For each Quadrant (cspad_layout.txt):

  Code Block
  none none
  def get_quad_image( self, data3d, qn) : """get_quad_image Get an image for this quad (qn) @param data3d 3d data array (row vs. col vs. section) @param qn quad number """ pairs = [] for i in range (8) : # 1) insert gap between asics in the 2x1 asics = np.hsplit( data3d[i], 2) gap = np.zeros( (185,3), dtype=data3d.dtype ) # # gap should be 3 pixels wide pair = np.hstack( (asics[0], gap, asics[1]) ) # all sections are originally 185 (rows) x 388 (columns) # Re-orient each section in the quad if i==0 or i==1 : pair = pair[:,::-1].T # reverse columns, switch columns to rows. if i==4 or i==5 : pair = pair[::-1,:].T # reverse rows, switch rows to columns pairs.append( pair ) if self.small_angle_tilt : pair = scipy.ndimage.interpolation.rotate(pair,self.tilt_array[qn][i]) # make the array for this quadrant quadrant = np.zeros( (self.npix_quad850, self.npix_quad850), dtype=data3d.dtype ) # insert the 2x1 sections according to for sec in range (8): nrows, ncols = pairs[sec].shape # colp,rowp are where the top-left corner of a section should be placed rowp = self.npix_quad850 - self.sec_offset[0] - (self.section_centers[0][qn][sec] + nrows/2) colp = self.npix_quad850 - self.sec_offset[1] - (self.section_centers[1][qn][sec] + ncols/2) quadrant[rowp:rowp+nrows, colp:colp+ncols] = pairs[sec][0:nrows,0:ncols] if (rowp+nrows > self.npix_quad) or (colp+ncols > self.npix_quad) : ) print "ERROR" print rowp, ":", quadrant[rowp:rowp+nrows, ", ", colp, ":",colp+ncolscolp:colp+ncols] = pairs[sec][0:nrows,0:ncols] return quadrant

Then combine all four quadrant images into the full detector image:

        self.image = np.zeros((2*self.npix_quad850+100, 2*self.npix_quad850+100 ), dtype="float64")
        for quad in xrange (4):

            quad_image = self.get_quad_image( self.pixels[quad], quad )
            self.qimages[quad] = quad_image

            if quad>0:
                # reorient the quad_image as needed
                quad_image = np.rot90( quad_image, 4-quad)

            qoff_x = self.quad_offset[0,quad]
            qoff_y = self.quad_offset[1,quad]
            self.image[qoff_x:qoff_x+self.npix_quad850, qoff_y:qoff_y+self.npix_quad850]=quad_image

        return self.image

...

Notice that the code snippets above make use of some predefined quantities which it reads in from "calibration files". The files contains calibrated numerical values for individual sections' and quads' rotations and shifts. All of these files are located in the experiment's 'calib' folder, but is not generated automatically. The XtcExplorer currently has a local version which is not correct but which is close enough to display a reasonable image. For how the files have been read in, you can take a look at XtcExplorer/src/cspad.py's read_alignment function.

For how to find the correct constants for each experiment, look at the CSPad alignment page.

Saving data arrays

saving numpy arrays to numpy file

...