def plot_keypoint_scales(hs, fnum=1):
print('[dev] plot_keypoint_scales()')
cx2_kpts = hs.feats.cx2_kpts
cx2_nFeats = map(len, cx2_kpts)
kpts = np.vstack(cx2_kpts)
print('[dev] --- LaTeX --- ')
_printopts = np.get_printoptions()
np.set_printoptions(precision=3)
print(pytex.latex_scalar(r'\# keypoints, ', len(kpts)))
print(pytex.latex_mystats(r'\# keypoints per image', cx2_nFeats))
acd = kpts[:, 2:5].T
scales = np.sqrt(acd[0] * acd[2])
scales = np.array(sorted(scales))
print(pytex.latex_mystats(r'keypoint scale', scales))
np.set_printoptions(**_printopts)
print('[dev] ---/LaTeX --- ')
#
df2.figure(fnum=fnum, docla=True, title='sorted scales')
df2.plot(scales)
df2.adjust_subplots_safe()
#ax = df2.gca()
#ax.set_yscale('log')
#ax.set_xscale('log')
#
fnum += 1
df2.figure(fnum=fnum, docla=True, title='hist scales')
df2.show_histogram(scales, bins=20)
df2.adjust_subplots_safe()
#ax = df2.gca()
#ax.set_yscale('log')
#ax.set_xscale('log')
return fnum
def plot_keypoint_scales(hs, fnum=1):
print('[dev] plot_keypoint_scales()')
cx2_kpts = hs.feats.cx2_kpts
cx2_nFeats = map(len, cx2_kpts)
kpts = np.vstack(cx2_kpts)
print('[dev] --- LaTeX --- ')
_printopts = np.get_printoptions()
np.set_printoptions(precision=3)
print(pytex.latex_scalar(r'\# keypoints, ', len(kpts)))
print(pytex.latex_mystats(r'\# keypoints per image', cx2_nFeats))
acd = kpts[:, 2:5].T
scales = np.sqrt(acd[0] * acd[2])
scales = np.array(sorted(scales))
print(pytex.latex_mystats(r'keypoint scale', scales))
np.set_printoptions(**_printopts)
print('[dev] ---/LaTeX --- ')
#
df2.figure(fnum=fnum, docla=True, title='sorted scales')
df2.plot(scales)
df2.adjust_subplots_safe()
#ax = df2.gca()
#ax.set_yscale('log')
#ax.set_xscale('log')
#
fnum += 1
df2.figure(fnum=fnum, docla=True, title='hist scales')
df2.show_histogram(scales, bins=20)
df2.adjust_subplots_safe()
#ax = df2.gca()
#ax.set_yscale('log')
#ax.set_xscale('log')
return fnum
def viz_localmax(signal1d):
#signal1d = np.array(hist)
from hsviz import draw_func2 as df2
signal1d = np.array(signal1d)
maxpos = np.array(localmax(signal1d))
x_data = range(len(signal1d))
y_data = signal1d
df2.figure('localmax vizualization')
df2.plot(x_data, y_data)
df2.plot(maxpos, signal1d[maxpos], 'ro')
df2.update()
def viz_localmax(signal1d):
#signal1d = np.array(hist)
from hsviz import draw_func2 as df2
signal1d = np.array(signal1d)
maxpos = np.array(localmax(signal1d))
x_data = range(len(signal1d))
y_data = signal1d
df2.figure('localmax vizualization')
df2.plot(x_data, y_data)
df2.plot(maxpos, signal1d[maxpos], 'ro')
df2.update()
def top_matching_features(res, axnum=None, match_type=''):
cx2_fs = res.cx2_fs_V
cx_fx_fs_list = []
for cx in xrange(len(cx2_fs)):
fx2_fs = cx2_fs[cx]
for fx in xrange(len(fx2_fs)):
fs = fx2_fs[fx]
cx_fx_fs_list.append((cx, fx, fs))
cx_fx_fs_sorted = np.array(sorted(cx_fx_fs_list, key=lambda x: x[2])[::-1])
sorted_score = cx_fx_fs_sorted[:, 2]
df2.figure(0)
df2.plot(sorted_score)
def top_matching_features(res, axnum=None, match_type=''):
cx2_fs = res.cx2_fs_V
cx_fx_fs_list = []
for cx in xrange(len(cx2_fs)):
fx2_fs = cx2_fs[cx]
for fx in xrange(len(fx2_fs)):
fs = fx2_fs[fx]
cx_fx_fs_list.append((cx, fx, fs))
cx_fx_fs_sorted = np.array(sorted(cx_fx_fs_list, key=lambda x: x[2])[::-1])
sorted_score = cx_fx_fs_sorted[:, 2]
df2.figure(0)
df2.plot(sorted_score)
