outlier_point_removal.py

# Given a set of ML pruned saddle points, remove outliers or keep only chessboard area
import numpy as np
import matplotlib.pyplot as plt


pts = np.array([[237, 332],
 [242, 287],
 [245, 263],
 [247, 360],
 [249, 337],
 [252, 314],
 [255, 290],
 [257, 389],
 [258, 266],
 [260, 366],
 [262, 342],
 [268, 294],
 [271, 269],
 [274, 372],
 [280, 323],
 [286, 272],
 [286, 404],
 [291, 353],
 [295, 328],
 [298, 302],
 [298, 437],
 [301, 411],
 [304, 386],
 [311, 333],
 [318, 279],
 [318, 420],
 [321, 393],
 [324, 366],
 [329, 339],
 [332, 311],
 [336, 283]])

outliers = np.array([
  [50,23],
  [30,63],
  [400,203],
  [250,370],
  [500,303],
  ])

pts = np.vstack([pts, outliers])


# plt.plot(outliers[:,0],outliers[:,1],'rx')
# plt.plot(pts[:,0],pts[:,1],'.')
# plt.show()
N = len(pts)

new_order = np.arange(N)
np.random.shuffle(new_order)
pts = pts[new_order,:]

# TODO Calculate closest N points over threshold instead of just closest
# because currently a pair of outliers next to each other are kept
# def calculateOutliers(pts, threshold_mult = 3):
#   N = len(pts)
#   dists = np.zeros([N,N])
#   best_dists = np.zeros(N)
#   for i in range(N):
#     dists[i,:] = np.linalg.norm(pts[:,:] - pts[i,:], axis=1)
#     x = np.linalg.norm(pts - pts[i,:], axis=1)
#     best_dists[i] = np.min(x[x!=0])
#   print(best_dists)
#   # med = np.median(dists[dists!=0])
#   med = np.median(best_dists)
#   print(med)
#   # return dists.min(axis=0) > med*threshold_mult
#   return best_dists > med*threshold_mult

def calculateOutliers(pts, threshold_mult = 1.5):
  N = len(pts)
  std = np.std(pts, axis=0)
  ctr = np.mean(pts, axis=0)
  return (np.any(np.abs(pts-ctr) > threshold_mult * std, axis=1))



import time

ta = time.time()

outlier = calculateOutliers(pts)
pred_outliers = pts[outlier,:]

tb =  time.time()

proc_time = tb-ta
print("Processed in %.2f ms" % (proc_time*1e3))


# plt.plot(outliers[:,0],outliers[:,1],'rx')
plt.plot(pred_outliers[:,0],pred_outliers[:,1],'rx')
plt.plot(pts[:,0],pts[:,1],'.')
plt.show()