def calculate_distance_matrix(app, all_imgs): hist_img = read_hist_img(app) distance_matrix = [([0] * len(all_imgs)) for i in range(len(all_imgs))] for i in range(len(all_imgs)): for j in range(len(all_imgs)): img_i = all_imgs[i] img_j = all_imgs[j] distance_matrix[i][j] = distance_euclidean_img(app, hist_img, img_i, img_j) return distance_matrix
def distance_img(app, img_name_a, img_name_b):
hist_img = read_hist_img(app)
index_a = get_img_pos(img_name_a)
index_b = get_img_pos(img_name_b)
img_a = hist_img[index_a]
img_b = hist_img[index_b]
dis = distance.euclidean(img_a, img_b)
return dis
def is_same_img(app, img_name_a, img_name_b): hist_img = read_hist_img(app) index_a = get_img_pos(img_name_a) index_b = get_img_pos(img_name_b) img_a = hist_img[index_a] img_b = hist_img[index_b] dis = distance.euclidean(img_a, img_b) if dis <= img_similar_threshold: return True return False
def calculate_distance(app):
if not os.path.exists('/'.join([DISTANCE_BASE_PATH, app])):
os.makedirs('/'.join([DISTANCE_BASE_PATH, app]))
all_reports = get_all_reports(app)
hist_txt = read_hist_txt(app)
hist_img = read_hist_img(app)
distance_matrix_txt = [([0] * len(all_reports))
for i in range(len(all_reports))]
distance_matrix_img = [([0] * len(all_reports))
for i in range(len(all_reports))]
distance_matrix_report = [([0] * len(all_reports))
for i in range(len(all_reports))]
# preprocess the hist in a dict to improve the efficiency
hist_txt_dict = {}
hist_img_dict = {}
for i in range(len(all_reports)):
report = all_reports[i]
hist_txt_dict[report] = get_hist_txt(app, hist_txt, report)
hist_img_dict[report] = get_hist_img(app, hist_img, report)
for i in range(len(all_reports)):
for j in range(len(all_reports)):
report_a = all_reports[i]
report_b = all_reports[j]
hist_txt_a = hist_txt_dict[report_a]
hist_txt_b = hist_txt_dict[report_b]
hist_img_a = hist_img_dict[report_a]
hist_img_b = hist_img_dict[report_b]
distance_txt = distance_txt_jaccard(hist_txt_a, hist_txt_b)
distance_imgset = distance_imgset_jaccard(hist_img_a, hist_img_b)
distance_report = distance_report_harmonic(distance_txt,
distance_imgset,
hist_img_a, hist_img_b)
distance_matrix_txt[i][j] = distance_txt
distance_matrix_img[i][j] = distance_imgset
distance_matrix_report[i][j] = distance_report
# save all distance
np.save('/'.join([DISTANCE_BASE_PATH, app, 'distance_txt.npy']),
distance_matrix_txt)
np.save('/'.join([DISTANCE_BASE_PATH, app, 'distance_img.npy']),
distance_matrix_img)
np.save('/'.join([DISTANCE_BASE_PATH, app, 'distance_harmonic.npy']),
distance_matrix_report)
def calculate_distance(app):
if not os.path.exists('/'.join([DISTANCE_BASE_PATH, app])):
os.makedirs('/'.join([DISTANCE_BASE_PATH, app]))
all_reports = get_all_reports(app)
hist_txt = read_hist_txt(app)
hist_img = read_hist_img(app)
distance_matrix_txt = [([0] * len(all_reports))
for i in range(len(all_reports))]
distance_matrix_img = [([0] * len(all_reports))
for i in range(len(all_reports))]
distance_matrix_report = [([0] * len(all_reports))
for i in range(len(all_reports))]
for i in range(len(all_reports)):
for j in range(len(all_reports)):
report_a = all_reports[i]
report_b = all_reports[j]
hist_txt_a = get_hist_txt(app, hist_txt, report_a)
hist_txt_b = get_hist_txt(app, hist_txt, report_b)
hist_img_a = get_hist_img(app, hist_img, report_a)
hist_img_b = get_hist_img(app, hist_img, report_b)
distance_txt = distance_txt_jaccard(hist_txt_a, hist_txt_b)
distance_imgset = distance_imgset_jaccard(hist_img_a, hist_img_b)
distance_report = distance_report_harmonic(distance_txt,
distance_imgset,
hist_img_a, hist_img_b)
distance_matrix_txt[i][j] = distance_txt
distance_matrix_img[i][j] = distance_imgset
distance_matrix_report[i][j] = distance_report
# save all distance
np.save('/'.join([DISTANCE_BASE_PATH, app, 'distance_txt.npy']),
distance_matrix_txt)
np.save('/'.join([DISTANCE_BASE_PATH, app, 'distance_img.npy']),
distance_matrix_img)
np.save('/'.join([DISTANCE_BASE_PATH, app, 'distance_harmonic.npy']),
distance_matrix_report)