基于极坐标变换的脱机手写藏文字符特征提取方法

摘要

为了使得藏文字符特征向量维数少、存储空间小、运算速度快及区分相似字能力高,基于图像投影法提出一种基于极坐标投影变换的脱机手写藏文字符特征提取方法.将脱机手写藏文字符图像进行预处理后得到大小、位置统一的二值图像,并定位二值图像的极点;求出二值图像中所有值为1的点对应的极坐标后将其进行投影变换得到投影向量,即作为脱机手写藏文字符的特征向量.使用KNN分类器对30000个脱机手写藏文字进行实验,其中80%的样本作为训练数据,20%的样本作为测试数据,识别率达到了96.32%.结果表明该方法的有效性、计算简单及达到了较好的识别效果.%A feature extraction method of off-line handwritten Tibetan characters based on projection transformation of polar coordinates is proposed to make less dimensions of Tibetan characters feature vector, less storage space, faster computation speed and higher ability of distinguishing similar words.The offline handwritten Tibetan character image is pre-processed to obtain binary image with same size and position, and its vertex is located.We obtain the polar coordinates corresponding to the points valued 1 in the binary image and the projected vector after projection transformation,which is feature vector for offline handwritten Tibetan characters.30 000 offline handwritten Tibetan characters are tested by using KNN classifier,80%of the samples are used as training data and 20%of the samples are used as test data.The recognition rate is 96.32% which indicates that the method is effective, simple and achieves better recognition results.