摘要
字符识别作为模式识别领域的一个重要分支,关键在于特征向量的选择和提取.本文在结合提升小波和分形二者特点的基础上提出了一种新的特征提取方法,实现了由二维图像数据向一维数据的转化.通过计算相应提升小波变换曲线的分形维数,得到新的特征向量.相关实验表明,本文算法是有效的,结果令人满意.
Character recognition is an important branch of pattern recognition. Its key point is the selecting and extracting proper feature vector. In this paper, a new way to get feature vector based on lifting wavelet and fractal dimension is proposed. It transforms 2-D image data to 1-D data which is decomposed by lifting wavelet. Afterwards the new feature vector is formed by calculating fractal dimensions of several curves. The result of experiment shows that this method is satisfied.
出处
《吉林化工学院学报》
CAS
2012年第11期97-99,共3页
Journal of Jilin Institute of Chemical Technology
关键词
特征提取
字符识别
提升小波
轮廓追踪
feature extraction
character recognition
liftingwavelet
contour pursuit
