def testdata_matching_affine_inliers_normalized():
tup = testdata_matching_affine_inliers()
kpts1, kpts2, fm, aff_inliers, rchip1, rchip2, xy_thresh_sqrd = tup
kpts1_ma = kpts1.take(fm.T[0].take(aff_inliers), axis=0)
kpts2_ma = kpts2.take(fm.T[1].take(aff_inliers), axis=0)
#kpts1_ma, kpts2_ma, rchip1, rchip2, xy_thresh_sqrd = testdata_matching_affine_inliers()
# Matching affine inliers
xy1_ma = ktool.get_xys(kpts1_ma)
xy2_ma = ktool.get_xys(kpts2_ma)
# Matching affine inliers normalized
xy1_man, T1 = ltool.whiten_xy_points(xy1_ma)
xy2_man, T2 = ltool.whiten_xy_points(xy2_ma)
return xy1_man, xy2_man, rchip1, rchip2, T1, T2
def testdata_matching_affine_inliers_normalized():
tup = testdata_matching_affine_inliers()
kpts1, kpts2, fm, aff_inliers, rchip1, rchip2, xy_thresh_sqrd = tup
kpts1_ma = kpts1.take(fm.T[0].take(aff_inliers), axis=0)
kpts2_ma = kpts2.take(fm.T[1].take(aff_inliers), axis=0)
#kpts1_ma, kpts2_ma, rchip1, rchip2, xy_thresh_sqrd = testdata_matching_affine_inliers()
# Matching affine inliers
xy1_ma = ktool.get_xys(kpts1_ma)
xy2_ma = ktool.get_xys(kpts2_ma)
# Matching affine inliers normalized
xy1_man, T1 = ltool.whiten_xy_points(xy1_ma)
xy2_man, T2 = ltool.whiten_xy_points(xy2_ma)
return xy1_man, xy2_man, rchip1, rchip2, T1, T2
def get_normalized_affine_inliers(kpts1, kpts2, fm, aff_inliers):
"""
returns xy-inliers that are normalized to have a mean of 0 and std of 1 as
well as the transformations so the inverse can be taken
"""
fm_affine = fm.take(aff_inliers, axis=0)
# Get corresponding points and shapes
kpts1_ma = kpts1.take(fm_affine.T[0], axis=0)
kpts2_ma = kpts2.take(fm_affine.T[1], axis=0)
#kpts1_ma = kpts1.take(fm_affine.T[0], axis=0)
#kpts2_ma = kpts2.take(fm_affine.T[1], axis=0)
# Normalize affine inliers xy locations
xy1_ma = ktool.get_xys(kpts1_ma)
xy2_ma = ktool.get_xys(kpts2_ma)
xy1_man, T1 = ltool.whiten_xy_points(xy1_ma)
xy2_man, T2 = ltool.whiten_xy_points(xy2_ma)
return xy1_man, xy2_man, T1, T2
def get_normalized_affine_inliers(kpts1, kpts2, fm, aff_inliers):
"""
returns xy-inliers that are normalized to have a mean of 0 and std of 1 as
well as the transformations so the inverse can be taken
"""
fm_affine = fm.take(aff_inliers, axis=0)
# Get corresponding points and shapes
kpts1_ma = kpts1.take(fm_affine.T[0], axis=0)
kpts2_ma = kpts2.take(fm_affine.T[1], axis=0)
#kpts1_ma = kpts1.take(fm_affine.T[0], axis=0)
#kpts2_ma = kpts2.take(fm_affine.T[1], axis=0)
# Normalize affine inliers xy locations
xy1_ma = ktool.get_xys(kpts1_ma)
xy2_ma = ktool.get_xys(kpts2_ma)
xy1_man, T1 = ltool.whiten_xy_points(xy1_ma)
xy2_man, T2 = ltool.whiten_xy_points(xy2_ma)
return xy1_man, xy2_man, T1, T2
