DEVELOPMENT OF AN EFFICIENT METHOD FOR DETECTING RECTANGULAR COMPONENTS FROM BINARY EDGE IMAGES : Application to architectural images
Haruka AMANUMA, Naoki Katoh · Journal of Environmental Engineering (Transactions of AIJ) · 2004
this paper proposes an efflcient algorithm fbr d¢ tecting rectangular componerits in a binary edge image by Hmiting targct ohlocts to 血ose having so -oallcd slicing struc 加 re ， The proposed algorithm first identifies the regions where ractangu 且 ar objectS may exist by pr｛ ) jecting the image intensity data to two axes ， It then fUnher narrows the set of candidate edges that are supposed te fbrm a part of rectangles by applying an efficient dyn 副 c prograrnming algorithm ． It has b¢ en demonStrated through computational experiments that our algori 血 m ε 飆 ns much highe π ef罰ciency and detection accu πacy compared with an existing method ， Kegwords ： binatJ ・ e4ge image， rectangle detection ， computer vision ， ． dyeiamic programming ， algerithm