An Approach of Removing Stroke Based on Touching Pattern for Handwritten Digits Segmentation

Abstract

-For finding precise segmentation, a new approach to remove and amend stroke connecting touching handwriting digits based on touching pattern is presented. Firstly, background region of touching digits is analyzed by using water reservoirs concept, and then the classification scheme is developed to identify the pattern of touching digits according to the structural features. In single-point touching type, two touching pairs of digits can be connected with a ligature, which can be a component of one of the digits or belong to no digit. After segmentation is performed to get isolated digits. the pattern-oriented ligature analysis investigates whether candidate ligature is a part of digit or a useless stroke and decides to be preserved or be removed. Experimental results show that the proposed approach is effective for segmentation of handwritten touching digits and segmentation errors have been reduced.