You should usually invoke this program as ocropus-hocr 'book/????.bin.png' For each page like 'book/0001.bin.png', it uses the following files: book/0001.bin.png # page image book/0001.pseg.png …

This is a quick and dirty implementation that works for up to a few thousand elements.""" n = len (order) visited = np.zeros (n) L = [] def visit (k): if visited [k]: return visited [k] = 1 for l in find (order [:,k]): visit …