def search_image(imagename):
with open(VOC_FILE, 'rb') as f:
voc = cPickle.load(f)
locs, descr = sift.read_features_from_file(get_sift_filename(imagename))
iw = voc.project(descr)
src = Searcher(DB_FILE, voc)
print 'ask using a histogram ...'
print src.candidates_from_histogram(iw)[:5]
print 'try a query ...'
for dist, imid in src.query(imagename)[:5]:
cur = src.con.execute('select filename from imlist where rowid=%d' % imid)
print '%s(%d)' % (cur.fetchone()[0], dist)
def index_images(imlist):
featlist = [ get_sift_filename(i) for i in imlist ]
# load vocabulary
with open(VOC_FILE, 'rb') as f:
voc = cPickle.load(f)
# create indexer
indx = Indexer(DB_FILE, voc)
indx.create_tables()
# go through all images, project features on vocabulary and insert
for imagefile, siftfile in zip(imlist, featlist):
locs, descr = sift.read_features_from_file(siftfile)
indx.add_to_index(imagefile, descr)
indx.db_commit()