def get_distances(self, vectors, image_dict):
results = {}
result = []
for i in vectors:
print("Working on hash" + str(i))
dist = {}
projection = []
for image in image_dict:
a = np.array(vectors[i], dtype=np.float)
b = a[1]
a = a[0]
p = np.array(image_dict[image], dtype=np.float)
ap = p - a
ab = b - a
# distance = np.array(np.around((ab * ap) / Distance.E2_distance(a, b), decimals=4))
projection.append(
np.array(
np.around((a + np.dot(ab, ap) / np.dot(ab, ab) * ab),
decimals=3)))
# projection.append(np.array(np.around()))
min_proj = np.amin(projection, axis=0)
image_keys = list(image_dict.keys())
j = 0
# print(min_proj)
for x in projection:
distance = Distance.E2_distance(min_proj, x)
ind = image_keys[j]
dist[ind] = distance
j = j + 1
result.append(dist)
results['h' + str(i)] = result
result = []
# print(results)
return results
def get_neighbors(self, labelled_set, labels, imageInstance, k):
distances = []
j = 0
for x in range(len(labelled_set)):
# dist = Similarity.cos_similarity(imageInstance, labelled_set[x])
dist = Distance.E2_distance(imageInstance, labelled_set[x])
distances.append((labelled_set[x], labels[j], dist))
j += 1
distances.sort(key=operator.itemgetter(2), reverse=False)
neighbors = []
for x in range(k):
neighbors.append(distances[x][1])
# print("neighbors:" + str(neighbors))
return neighbors
