def ransac(data,
model_class,
min_samples,
residual_threshold,
is_data_valid=None,
is_model_valid=None,
max_trials=100,
stop_sample_num=np.inf,
stop_residuals_sum=0,
stop_probability=1,
random_state=None,
initial_inliers=None):
best_model = None
best_inlier_num = 0
best_inlier_residuals_sum = np.inf
best_inliers = None
random_state = check_random_state(random_state)
num_samples = len(data[0])
for num_trials in range(max_trials):
# choose random sample
spl_idxs = random_state.choice(num_samples, min_samples, replace=False)
samples = [d[spl_idxs] for d in data]
sample_model = model_class()
success = sample_model.estimate(samples)
sample_model_residuals = np.abs(sample_model.residuals(data))
sample_model_residuals_sum = np.sum(sample_model_residuals**2)
sample_model_inliers = sample_model_residuals < residual_threshold
sample_inlier_num = np.sum(sample_model_inliers)
if (sample_inlier_num > best_inlier_num or
(sample_inlier_num == best_inlier_num
and sample_model_residuals_sum < best_inlier_residuals_sum)):
best_model = sample_model
best_inlier_num = sample_inlier_num
best_inlier_residuals_sum = sample_model_residuals_sum
best_inliers = sample_model_inliers
dynamic_max_trials = _dynamic_max_trials(best_inlier_num,
num_samples, min_samples,
stop_probability)
if (best_inlier_num >= stop_sample_num
or best_inlier_residuals_sum <= stop_residuals_sum
or num_trials >= dynamic_max_trials):
break
# estimate final model using all inliers
if best_inliers is not None:
# select inliers for each data array
data_inliers = [d[best_inliers] for d in data]
best_model.estimate(data_inliers)
return best_model, best_inliers
def test_ransac_dynamic_max_trials():
# Numbers hand-calculated and confirmed on page 119 (Table 4.3) in
# Hartley, R.~I. and Zisserman, A., 2004,
# Multiple View Geometry in Computer Vision, Second Edition,
# Cambridge University Press, ISBN: 0521540518
# e = 0%, min_samples = X
assert_equal(_dynamic_max_trials(100, 100, 2, 0.99), 1)
# e = 5%, min_samples = 2
assert_equal(_dynamic_max_trials(95, 100, 2, 0.99), 2)
# e = 10%, min_samples = 2
assert_equal(_dynamic_max_trials(90, 100, 2, 0.99), 3)
# e = 30%, min_samples = 2
assert_equal(_dynamic_max_trials(70, 100, 2, 0.99), 7)
# e = 50%, min_samples = 2
assert_equal(_dynamic_max_trials(50, 100, 2, 0.99), 17)
# e = 5%, min_samples = 8
assert_equal(_dynamic_max_trials(95, 100, 8, 0.99), 5)
# e = 10%, min_samples = 8
assert_equal(_dynamic_max_trials(90, 100, 8, 0.99), 9)
# e = 30%, min_samples = 8
assert_equal(_dynamic_max_trials(70, 100, 8, 0.99), 78)
# e = 50%, min_samples = 8
assert_equal(_dynamic_max_trials(50, 100, 8, 0.99), 1177)
# e = 0%, min_samples = 5
assert_equal(_dynamic_max_trials(1, 100, 5, 0), 0)
assert_equal(_dynamic_max_trials(1, 100, 5, 1), np.inf)
def test_ransac_dynamic_max_trials():
# Numbers hand-calculated and confirmed on page 119 (Table 4.3) in
# Hartley, R.~I. and Zisserman, A., 2004,
# Multiple View Geometry in Computer Vision, Second Edition,
# Cambridge University Press, ISBN: 0521540518
# e = 0%, min_samples = X
assert_equal(_dynamic_max_trials(100, 100, 2, 0.99), 1)
# e = 5%, min_samples = 2
assert_equal(_dynamic_max_trials(95, 100, 2, 0.99), 2)
# e = 10%, min_samples = 2
assert_equal(_dynamic_max_trials(90, 100, 2, 0.99), 3)
# e = 30%, min_samples = 2
assert_equal(_dynamic_max_trials(70, 100, 2, 0.99), 7)
# e = 50%, min_samples = 2
assert_equal(_dynamic_max_trials(50, 100, 2, 0.99), 17)
# e = 5%, min_samples = 8
assert_equal(_dynamic_max_trials(95, 100, 8, 0.99), 5)
# e = 10%, min_samples = 8
assert_equal(_dynamic_max_trials(90, 100, 8, 0.99), 9)
# e = 30%, min_samples = 8
assert_equal(_dynamic_max_trials(70, 100, 8, 0.99), 78)
# e = 50%, min_samples = 8
assert_equal(_dynamic_max_trials(50, 100, 8, 0.99), 1177)
# e = 0%, min_samples = 5
assert_equal(_dynamic_max_trials(1, 100, 5, 0), 0)
assert_equal(_dynamic_max_trials(1, 100, 5, 1), np.inf)
