def test_extract_region():
R, C = 4, 6
x = np.arange(R*C)
x.shape = (R, C)
for r in range(R):
for c in range(C):
s = util.extract_region_safe(x, r, c, 0, 17)
assert_equal(s.shape, (1, 1))
assert_equal(s[0, 0], x[r, c])
# upper left
s = util.extract_region_safe(x, 0, 0, 1, 17)
assert_equal(s.shape, (3, 3))
assert_array_equal(s[0, :], 17)
assert_array_equal(s[:, 0], 17)
assert_array_equal(s[1:,1:], x[:2, :2])
# upper right
s = util.extract_region_safe(x, 0, 5, 1, 17)
assert_equal(s.shape, (3, 3))
assert_array_equal(s[0, :], 17)
assert_array_equal(s[:, 2], 17)
assert_array_equal(s[1:,:2], x[:2, 4:])
# lower left
s = util.extract_region_safe(x, 3, 0, 1, 17)
assert_equal(s.shape, (3, 3))
assert_array_equal(s[2, :], 17)
assert_array_equal(s[:, 0], 17)
assert_array_equal(s[:2,1:], x[2:, :2])
# lower right
s = util.extract_region_safe(x, 3, 5, 1, 17)
assert_equal(s.shape, (3, 3))
assert_array_equal(s[2, :], 17)
assert_array_equal(s[:, 2], 17)
assert_array_equal(s[:2,:2], x[2:, 4:])
# now we move the window around for a variety of edge sizes,
# save the middle pixel, and reconstruct the matrix and compare
for p in range(4):
newx= np.zeros_like(x)
for r in range(R):
for c in range(C):
s = util.extract_region_safe(x, r, c, p, 17)
newx[r, c] = s[p, p]
assert_array_equal(newx, x)
def test_extract_region():
R, C = 4, 6
x = np.arange(R * C)
x.shape = (R, C)
for r in range(R):
for c in range(C):
s = util.extract_region_safe(x, r, c, 0, 17)
assert_equal(s.shape, (1, 1))
assert_equal(s[0, 0], x[r, c])
# upper left
s = util.extract_region_safe(x, 0, 0, 1, 17)
assert_equal(s.shape, (3, 3))
assert_array_equal(s[0, :], 17)
assert_array_equal(s[:, 0], 17)
assert_array_equal(s[1:, 1:], x[:2, :2])
# upper right
s = util.extract_region_safe(x, 0, 5, 1, 17)
assert_equal(s.shape, (3, 3))
assert_array_equal(s[0, :], 17)
assert_array_equal(s[:, 2], 17)
assert_array_equal(s[1:, :2], x[:2, 4:])
# lower left
s = util.extract_region_safe(x, 3, 0, 1, 17)
assert_equal(s.shape, (3, 3))
assert_array_equal(s[2, :], 17)
assert_array_equal(s[:, 0], 17)
assert_array_equal(s[:2, 1:], x[2:, :2])
# lower right
s = util.extract_region_safe(x, 3, 5, 1, 17)
assert_equal(s.shape, (3, 3))
assert_array_equal(s[2, :], 17)
assert_array_equal(s[:, 2], 17)
assert_array_equal(s[:2, :2], x[2:, 4:])
# now we move the window around for a variety of edge sizes,
# save the middle pixel, and reconstruct the matrix and compare
for p in range(4):
newx = np.zeros_like(x)
for r in range(R):
for c in range(C):
s = util.extract_region_safe(x, r, c, p, 17)
newx[r, c] = s[p, p]
assert_array_equal(newx, x)
PIX_REGION = 24 # num pixels on either side
ledimgs = np.zeros(
(len(valid_frames), 2, PIX_REGION * 2 + 1, PIX_REGION * 2 + 1),
dtype=np.uint8)
framepos = 0
for frame_chunk in util.chunk(valid_frames, 100):
frames = organizedata.get_frames(basedir, frame_chunk)
for frame_idx, frame in zip(frame_chunk, frames):
for led, field in [(0, 'led_front'), (1, 'led_back')]:
real_pos = positions_cleaned[field][frame_idx]
x, y = env.gc.real_to_image(real_pos[0], real_pos[1])
ledimgs[framepos, led, :, :] = util.extract_region_safe(
frame, int(y), int(x), PIX_REGION, 0)
framepos += 1
ledimgs_mean = np.mean(ledimgs.astype(np.float32), axis=0)
subsamp = 60
led_params_dict = {
'ledimgs_mean': ledimgs_mean,
'dist': compute_led_sep(positions[valid_frames]),
'subsampled_frames': ledimgs[::subsamp]
}
pickle.dump(led_params_dict, open(led_params, 'w'))
def led_measure(img_row):
PIX_REGION = 24 # num pixels on either side
ledimgs = np.zeros((len(valid_frames), 2, PIX_REGION*2+1,
PIX_REGION*2+1), dtype = np.uint8)
framepos = 0
for frame_chunk in util.chunk(valid_frames, 100):
frames = organizedata.get_frames(basedir, frame_chunk)
for frame_idx, frame in zip(frame_chunk, frames):
for led, field in [(0, 'led_front'),
(1, 'led_back')]:
real_pos = positions_cleaned[field][frame_idx]
x, y = env.gc.real_to_image(real_pos[0], real_pos[1])
ledimgs[framepos, led, :, :] = util.extract_region_safe(frame, int(y), int(x), PIX_REGION, 0)
framepos +=1
ledimgs_mean = np.mean(ledimgs.astype(np.float32),
axis=0)
subsamp = 60
led_params_dict = {'ledimgs_mean' : ledimgs_mean,
'dist' : compute_led_sep(positions[valid_frames]),
'subsampled_frames' : ledimgs[::subsamp]}
pickle.dump(led_params_dict, open(led_params, 'w'))
def led_measure(img_row):
import scipy.stats
midval = (np.max(img_row) - np.min(img_row)) / 2.0 + np.min(img_row)
