Curvature Based Analysis of Handwritten Symbols

Abstract

Pattern recognition is an important branch of computer vision and intelligent robotics. This paper explores local curve theory for different pattern recognition applications: recognition of good handwritten symbols (Fig 1 ), analysis of the hierarchical structure of 2-D patterns and simple 3-D object recognition. In the case of 2-D patterns, our technique is compared with a recent related method based on generalized Hough transforms and it is found that it is more robust with respect to complex image outlines.