摘要
针对变形字体的识别问题,提出了一种基于形状上下文和模糊推理相结合的度量机制的变形字体识别方法。首先,在分析现行字体识别方法的基础上,提出了薄板样条曲线TPS与模糊推理相结合的相似性距离度量机制的总体思路;其次,描述了基于形状上下文的变形字体图像的特征提取过程;而后讨论了薄板样条曲线TPS与模糊推理相结合的相似性距离度量机制,并将其应用在变形字体的识别算法中。实验结果表明所提出的方法是有效的。
A new text recognition algorithm is proposed based on Shape Context and Fuzzy inference to solve the corresponding problems. Firstly, existing deformation text recognition algorithms are analyzed, the main frame of distance measurement based on Thin Plate Spline （TPS） and Fuzzy inference is presented. Secondly, feature extraction based on Shape Context for deformation text is discussed. Lastly, the algorithm based on TPS and Fuzzy inference is used to improve accuracy and effectiveness of recognition. Then experiments results show that our proposed method is effective.
出处
《电子设计工程》
2014年第12期160-163,共4页
Electronic Design Engineering
基金
国家自然科学基金项目(60972095)
西安工业大学北方信息工程学院2013年度院长科研基金资助项目(YZ1310)
关键词
变形字体识别
形状上下文
薄板样条曲线
模糊推理
deformation text recognition
shape context
thin plate spline
fuzzy inference
