Low-Level Cursive Word Representation Based on Geometric Decomposition

Abstract. An efficient low-level word image representation plays a crucial role in general cursive word recognition. This paper proposes a novel representation scheme, where a word image can be represented as two sequences of feature vectors in two independent channels, which are extracted from vertical peak points on the upper external contour and at vertical minima on the lower external contour, respectively. A datadriven method based on support vector machine is applied to prune and group those extreme points. Our experimental results look promising and have indicated the potential of this low-level representation for complete cursive handwriting recognition.

• Type: Conference
• Year: 2005
• Where: MLDM
• Authors: Jian-xiong Dong, Adam Krzyzak, Ching Y. Suen, Dominique Ponson
• Keywords: Cursive Word Recognition, External Contour, Machine Learning, MLDM 2005, Word Image Representation