Algorithm of peak_finder_v3
- makes a map of pixels with local maximums of requested rank for data ndarray and mask, pixel code in the map may have bits 0/1/2/4 standing for not-a-maximum / maximum-in-row / maximum-in-column / maximum-in-rectangular-region of radius=rank.
- for each pixel with local maximal intensity in the region defined by the rank radius counts a number of pixels with intensity above zero, total positive intensity, center of gravity coordinates and rms,
- using parameters
r0(ex.=5.0), dr(ex.=0.05)
evaluates background level, rms of noise, and S/N for the pixel with maximal intensity. - makes list of peaks which pass selector with parameters set in
set_peak_selection_pars
, for example
alg.set_peak_selection_pars(npix_min=5, npix_max=500, amax_thr=0, atot_thr=1000, son_min=6)
Demonstration for local maximum map
Test for 100x100 image with random normal distribution of intensities
Example of the map of local maximums found for rank from 1 to 5:
color coding of pixels:
- blue=0 - not a local maximum
- green=1 - local maximum in row
- yellow=1+2 - local maximum in row and column
- red=1+2+4 - local maximum in rectangular region of radius=rank.
Table for rank, associated 2-d region size, fraction of pixels recognized as local maximums for rank, and time consumption for this algorithm.
rank | 2-d region | fraction | time, ms |
---|---|---|---|
1 | 3x3 | 0.1062 | 5.4 |
2 | 5x5 | 0.0372 | 5.2 |
3 | 7x7 | 0.0179 | 5.1 |
4 | 9x9 | 0.0104 | 5.2 |
5 | 11x11 | 0.0066 | 5.2 |
Overview
Content Tools