计算机技术与发展2016，Vol.26Issue(3)：93-96,4.DOI:10.3969/j.issn.1673-629X.2016.03.022
基于多投影的脱机手写藏文字符特征提取方法
Feature Extraction Method of Off-line Handwritten Tibetan Character Based on Multiple Projection
摘要
Abstract
Off-line handwritten Tibetan character recognition can promote the development and propagation of Tibetan culture,and the method of classification is based on the feature of off-line handwritten Tibetan character image. Since original definition of projection en-tropy does not make full use of image information,a feature extraction method of off-line handwritten Tibetan character based on multiple projection normalization is proposed. Firstly,an off-line handwritten Tibetan character image is scanned in horizontal,vertical,main diag-onal and secondary diagonal directions to create a row projection vector,a column projection vector,a main projection vector and a sec-ondary projection vector. Secondly,all projection vectors are normalized to create a multiple projection normalized vector that is the fea-ture vector for this character image. Finally,KNN classifier is used in classification. The proposed method is tested on the developed off-line Tibetan handwritten character sample database. The experimental results demonstrate that the proposed method is not only easy in cal-culation but also efficient in recognition accuracy.关键词
脱机手写藏文字符/特征提取/投影向量/归一化/多投影归一化向量Key words
off-line handwritten Tibetan character/feature extraction/projection vector/normalization/multiple projection normalized vector分类
计算机与自动化引用本文复制引用
蔡晓娟,黄鹤鸣..基于多投影的脱机手写藏文字符特征提取方法[J].计算机技术与发展,2016,26(3):93-96,4.基金项目
国家自然科学基金资助项目(61462072) （61462072）
教育部春晖计划项目(Z2104020) （Z2104020）
