This paper proposes a novel method in detecting the four reference lin
es of a cursive word at any skew angle. The proposed method is rotatio
n independent because the detection is based on points determined by t
heir relative positions between the outer contour of a cursive word an
d its convex hull. Features like bounding box, contour area, convex hu
ll and furthest points are used in detecting and grouping these points
. The reference lines are detected by searching for suitable peaks in
the Hough transformation space of these points. Encouraging experiment
al results have been obtained. (C) 1997 Pattern Recognition Society.