摘要
纸质图书和书法书籍的数字化及网络传播,方便书法学术的研究和书法爱好者的使用。为了对书法风格进行识别,提出基于特征书法风格分类方法:首先,对单字图像进行特征提取。接着,用爬虫技术,将单字转化为笔画,提取笔画特征。然后,将提取的24类特征作为特征向量,构造风格模型。最后,提取用户提交样本图的风格特征,与五类风格进行相似性比较,将样本字归属为概率最大的风格类型。
The digitalization of calligraphy paper books enables convenient use for academic researchers and calligraphy learners. Identifies the cal-ligraphy style by extracting and modeling calligraphy image features in character level and stroke level： Firstly extracts characters features. Second, extracts features of stroke by the crawler and stroke features are extracted. Totally, 24 style features are used as the feature vector, when a user submits an unknown character, its 24 style features are extracted and compared with those features of 5 styles in the database five styles eventually, the style which has the biggest similarity probability assigned to the unknown.
出处
《现代计算机》
2016年第14期39-46,共8页
Modern Computer
关键词
书法风格
风格识别
视觉特征
风格量化
Calligraphy Style
Style Identification
Visual Features
Style Quantification
