File size: 1,677 Bytes
2a3480c | 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 45 46 47 48 49 50 51 52 53 54 | """Reference repair for validation only; excluded from training images."""
from pathlib import Path
Path('libfmp/c3/c3s2_dtw.py').write_text('''import numpy as np
from scipy.spatial.distance import cdist
def compute_cost_matrix(X, Y, metric="euclidean"):
return cdist(X.T, Y.T, metric=metric)
def _accumulate(C, steps):
D = np.full(C.shape, np.inf)
D[0, 0] = C[0, 0]
for i in range(C.shape[0]):
for j in range(C.shape[1]):
if i == j == 0:
continue
previous = [D[i-di, j-dj] for di, dj in steps if i >= di and j >= dj]
if previous:
D[i, j] = C[i, j] + min(previous)
return D
def _path(D, steps):
i, j = D.shape[0]-1, D.shape[1]-1
points = [(i, j)]
while i or j:
candidates = [(i-di, j-dj) for di, dj in steps if i >= di and j >= dj]
if not candidates:
raise ValueError("No feasible warping path")
i, j = min(candidates, key=lambda p: D[p])
points.append((i, j))
return np.array(points[::-1])
def compute_accumulated_cost_matrix(C):
return _accumulate(C, [(1, 1), (1, 0), (0, 1)])
def compute_optimal_warping_path(D):
return _path(D, [(1, 1), (1, 0), (0, 1)])
def compute_accumulated_cost_matrix_21(C):
return _accumulate(C, [(1, 1), (2, 1), (1, 2)])
def compute_optimal_warping_path_21(D):
return _path(D, [(1, 1), (2, 1), (1, 2)])
''')
with Path('libfmp/c3/__init__.py').open('a') as output:
output.write('\nfrom .c3s2_dtw import compute_cost_matrix, compute_accumulated_cost_matrix, compute_optimal_warping_path, compute_accumulated_cost_matrix_21, compute_optimal_warping_path_21\n')
|