Project

General

Profile

Actions

Feature #83

closed

Skewness of particle intesity

Added by Neil Voss almost 15 years ago. Updated almost 15 years ago.

Status:
Closed
Priority:
Low
Assignee:
Amber Herold
Category:
-
Target version:
-
Start date:
02/11/2010
Due date:
% Done:

0%

Estimated time:
1.00 h
Spent time:

Description

Cathy Lawson found that the skewness of the particle intensity may be able to filter out particles over the carbon (at least carbon edges). Should we add this to makestack?

Actions #1

Updated by Neil Voss almost 15 years ago

  • Status changed from New to In Code Review
  • Assignee set to Amber Herold
  • Estimated time set to 1.00 h

added additional statistics in stack making, including:

  • min
  • max
  • skew
  • kurtosis
  • edgemean
  • edgestdev
  • centermean
  • centerstdev.

Tested on a couple of projects. Fixed in revision r13049. Had to fix a function r13048 in source:trunk/pyami/imagefun.py to make sure filled_circle actually used the center of an array.

Actions #2

Updated by Amber Herold almost 15 years ago

  • Assignee changed from Amber Herold to Neil Voss

Two things:
1. Is (box/2.0-1.0) supposed to be the radius of the shape? I don't understand what this calculates.
2. I like that the circle center calc is changed to a float, but why subtract .5?

I like it when magic numbers are explained in the code so I don't go messing it up after you are gone!

Actions #3

Updated by Neil Voss almost 15 years ago

  • Assignee changed from Neil Voss to Amber Herold

The 0.5 was really based on testing, but to get the the center of a NxN array where N is even the center is (N/2 - 1/2), the half pixel shift is because the array size is even, which required for EMAN. So the center of a 2x2 box is (0.5,0.5) the gap in between the pixels. For the odd array 3x3, the center is a whole pixel namely (1,1).

0,0 1,0
0,1 1,1 --> 0.5, 0.5 is center

0,0 1,0 2,0
0,1 1,1 2,1
0,2 1,2 2,2 --> 1,1 is center

You are right I should explain this in the code, but I have no SVN access at the moment.

Actions #4

Updated by Neil Voss almost 15 years ago

Oh yeah, box/2 - 1 is half box minus 1. I use less one for small boxes to make sure I have a sufficient number of pixels, but it really does not matter as long as we are consistent. I will document this when I return.

Actions #5

Updated by Amber Herold almost 15 years ago

  • Status changed from In Code Review to Closed

There is not a good way to test this other than examining the database. It provides stats that may be useful in the future.

Actions

Also available in: Atom PDF