lsst.meas.astrom  15.0-3-g52118bc
cleanBadPoints.py
Go to the documentation of this file.
1 #
2 # LSST Data Management System
3 # Copyright 2008, 2009, 2010 LSST Corporation.
4 #
5 # This product includes software developed by the
6 # LSST Project (http://www.lsst.org/).
7 #
8 # This program is free software: you can redistribute it and/or modify
9 # it under the terms of the GNU General Public License as published by
10 # the Free Software Foundation, either version 3 of the License, or
11 # (at your option) any later version.
12 #
13 # This program is distributed in the hope that it will be useful,
14 # but WITHOUT ANY WARRANTY; without even the implied warranty of
15 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 # GNU General Public License for more details.
17 #
18 # You should have received a copy of the LSST License Statement and
19 # the GNU General Public License along with this program. If not,
20 # see <http://www.lsstcorp.org/LegalNotices/>.
21 #
22 
23 __all__ = ["clean"]
24 
25 
26 import numpy as np
27 
28 from . import LeastSqFitter1dPoly
29 
30 
31 def clean(srcMatch, wcs, order=3, nsigma=3):
32  """Remove bad points from srcMatch
33 
34  Input:
35  srcMatch : list of det::SourceMatch
36  order: Order of polynomial to use in robust fitting
37  nsigma: Sources more than this far away from the robust best fit
38  polynomial are removed
39 
40  Return:
41  list of det::SourceMatch of the good data points
42  """
43 
44  N = len(srcMatch)
45  catX = np.zeros(N)
46  # catY = np.zeros(N)
47  for i in range(N):
48  x, y = wcs.skyToPixel(srcMatch[i].first.getCoord())
49  catX[i] = x
50  # catY[i] = y
51 
52  # TODO -- why does this only use X?
53 
54  x = np.array([s.second.getX() for s in srcMatch])
55  dx = x - catX
56  sigma = np.zeros_like(dx) + 0.1
57 
58  idx = indicesOfGoodPoints(x, dx, sigma, order=order, nsigma=nsigma)
59 
60  clean = []
61  for i in idx:
62  clean.append(srcMatch[i])
63  return clean
64 
65 
66 def indicesOfGoodPoints(x, y, s, order=1, nsigma=3, maxiter=100):
67  """Return a list of indices in the range [0, len(x)]
68  of points that lie less than nsigma away from the robust
69  best fit polynomial
70  """
71 
72  # Indices of elements of x sorted in order of increasing value
73  idx = x.argsort()
74  newidx = []
75  for niter in range(maxiter):
76  rx = chooseRx(x, idx, order)
77  ry = chooseRy(y, idx, order)
78  rs = np.ones((len(rx)))
79 
80  lsf = LeastSqFitter1dPoly(list(rx), list(ry), list(rs), order)
81  fit = [lsf.valueAt(value) for value in rx]
82  f = [lsf.valueAt(value) for value in x]
83 
84  sigma = (y-f).std()
85  deviance = np.fabs((y - f) / sigma)
86  newidx = np.flatnonzero(deviance < nsigma)
87 
88  if False:
89  import matplotlib.pyplot as plt
90  plt.plot(x, y, 'ks')
91  plt.plot(rx, ry, 'b-')
92  plt.plot(rx, ry, 'bs')
93  plt.plot(rx, fit, 'ms')
94  plt.plot(rx, fit, 'm-')
95  # plt.plot(x[newidx], y[newidx], 'rs')
96  plt.show()
97 
98  # If we haven't culled any points we're finished cleaning
99  if len(newidx) == len(idx):
100  break
101 
102  # We get here because we either a) stopped finding bad points
103  # or b) ran out of iterations. Either way, we just return our
104  # list of indices of good points.
105  return newidx
106 
107 
108 def chooseRx(x, idx, order):
109  """Create order+1 values of the ordinate based on the median of groups of elements of x"""
110  rSize = len(idx)/float(order+1) # Note, a floating point number
111  rx = np.zeros((order+1))
112 
113  for i in range(order+1):
114  rng = list(range(int(rSize*i), int(rSize*(i+1))))
115  rx[i] = np.mean(x[idx[rng]])
116  return rx
117 
118 
119 def chooseRy(y, idx, order):
120  """Create order+1 values of the ordinate based on the median of groups of elements of y"""
121  rSize = len(idx)/float(order+1) # Note, a floating point number
122  ry = np.zeros((order+1))
123 
124  for i in range(order+1):
125  rng = list(range(int(rSize*i), int(rSize*(i+1))))
126  ry[i] = np.median(y[idx[rng]])
127  return ry
STL namespace.
def indicesOfGoodPoints(x, y, s, order=1, nsigma=3, maxiter=100)
def clean(srcMatch, wcs, order=3, nsigma=3)