3DReconstruction/ransac.py at master · samehkhamis/3DReconstruction · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
from pylab import *

def ransac(model, x, y, nsamples, threshold, maxiter = 1e2, desiredprob = 0.99, debug = False):
	iterations = 0
	ndata = x.shape[0]
	best_inlier_idxs = None
	best_ninliers = 0
	goodprob = 0
	bestfit = None
	while goodprob < desiredprob and iterations < maxiter:
		all_idxs = arange(ndata)
		shuffle(all_idxs)
		maybe_idxs = all_idxs[:nsamples]
		test_idxs = all_idxs[nsamples:]

		x_maybeinliers = x[maybe_idxs,:]
		y_maybeinliers = y[maybe_idxs,:]
		x_test_points = x[test_idxs]
		y_test_points = y[test_idxs]

		maybemodel = model.fit(x_maybeinliers, y_maybeinliers)
		test_err = model.error(maybemodel, x_test_points, y_test_points)
		also_idxs = test_idxs[test_err < threshold] # select indices of rows with accepted points
		x_alsoinliers = x[also_idxs,:]
		y_alsoinliers = y[also_idxs,:]

		ninliers = also_idxs.shape[0]
		if ninliers > best_ninliers:
			betterx = concatenate((x_maybeinliers, x_alsoinliers))
			bettery = concatenate((y_maybeinliers, y_alsoinliers))
			bestfit = model.fit(betterx, bettery)
			best_inlier_idxs = concatenate((maybe_idxs, also_idxs))
			best_ninliers = ninliers

		iterations += 1
		goodprob = 1 - (1 - (float(best_ninliers) / ndata)**nsamples)**iterations
		if debug:
			print 'test_err.min()',test_err.min()
			print 'test_err.max()',test_err.max()
			print 'mean(test_err)',mean(test_err)
			print 'iteration %d: inliers = %d, goodprob = %f' % (iterations, ninliers, goodprob)

	return bestfit, best_inlier_idxs