def findWord(dict_vocab,path,grid_m,grid_n):
files = [ f for f in listdir(path) if isfile(join(path,f)) ]
word_hist = []
for f in files:
keypoints,file_desc = test_feature_detector(path+f, grid_n)
line_hist = []
for i in range(0,grid_m):
for j in range(0,grid_n):
desc = array(file_desc[j])
hist = buildWordHist(desc,dict_vocab[0])
# if len(line_hist) == 0:
# line_hist = hist
#else:
# line_hist = np.hstack((line_hist,hist))
line_hist.append(hist)
word_hist.append(line_hist)
return word_hist
def findWord(dict_vocab, path, grid_m, grid_n):
files = [f for f in listdir(path) if isfile(join(path, f))]
word_hist = []
for f in files:
keypoints, file_desc = test_feature_detector(path + f, grid_n)
line_hist = []
for i in range(0, grid_m):
for j in range(0, grid_n):
desc = array(file_desc[j])
hist = buildWordHist(desc, dict_vocab[0])
# if len(line_hist) == 0:
# line_hist = hist
#else:
# line_hist = np.hstack((line_hist,hist))
line_hist.append(hist)
word_hist.append(line_hist)
return word_hist
def buildVocabulary(path,k,grid_m,grid_n):
files = [ f for f in listdir(path) if isfile(join(path,f)) ]
total_desc = []
dict_vocab = []
database_keypoints = []
database_file_desc = []
for f in files:
print f
keypoints,file_desc = test_feature_detector(path+f, grid_n)
database_keypoints.append(keypoints)
database_file_desc.append(file_desc)
for i in range(0,grid_m):
for j in range(0,grid_n):
if len(total_desc) < 1:
total_desc.append(file_desc[j])
else:
temp = total_desc[0]
total_desc[0] = np.vstack((temp,file_desc[j]))
t1 = time.time()
vocab,dist = kmeans(total_desc[0],k) # k is the seed number
t2 = time.time()
print 'Kmeans in grid[',j,'] takes',t2-t1
dict_vocab.append(vocab)
word_hist = []
for fidx in range(0,len(files)):
keypoints = database_keypoints[fidx]
file_desc = database_file_desc[fidx]
line_hist = []
for i in range(0,grid_m):
for j in range(0,grid_n):
desc = array(file_desc[j])
hist = buildWordHist(desc,dict_vocab[0])
#if len(line_hist) == 0:
# line_hist = hist
#else:
# line_hist = np.hstack((line_hist,hist))
line_hist.append(hist)
word_hist.append(line_hist)
return dict_vocab,word_hist
def buildVocabulary(path, k, grid_m, grid_n):
files = [f for f in listdir(path) if isfile(join(path, f))]
total_desc = []
dict_vocab = []
database_keypoints = []
database_file_desc = []
for f in files:
print f
keypoints, file_desc = test_feature_detector(path + f, grid_n)
database_keypoints.append(keypoints)
database_file_desc.append(file_desc)
for i in range(0, grid_m):
for j in range(0, grid_n):
if len(total_desc) < 1:
total_desc.append(file_desc[j])
else:
temp = total_desc[0]
total_desc[0] = np.vstack((temp, file_desc[j]))
t1 = time.time()
vocab, dist = kmeans(total_desc[0], k) # k is the seed number
t2 = time.time()
print 'Kmeans in grid[', j, '] takes', t2 - t1
dict_vocab.append(vocab)
word_hist = []
for fidx in range(0, len(files)):
keypoints = database_keypoints[fidx]
file_desc = database_file_desc[fidx]
line_hist = []
for i in range(0, grid_m):
for j in range(0, grid_n):
desc = array(file_desc[j])
hist = buildWordHist(desc, dict_vocab[0])
#if len(line_hist) == 0:
# line_hist = hist
#else:
# line_hist = np.hstack((line_hist,hist))
line_hist.append(hist)
word_hist.append(line_hist)
return dict_vocab, word_hist