摘要
标牌压印字符是“反光差”的凹凸字符,通常的基于二值化图像的字符特征抽取方法都不适宜。提出了基于灰度图像的标牌压印字符特征抽取新方法,首先对灰度字符进行圆周投影,然后利用小波变换,将投影曲线分解为大致信号和细节信号的子样本,最后对子样本进行DCT变换,生成凹凸字符的特征矢量。该方法是直接对灰度图像抽取字符特征,不仅可以尽量多地保持原始字符的特征,而且克服了传统的字符图像特征抽取时,过分依赖于二值化算法、抗干扰性差等弊病。对标牌压印有限凹凸字符集进行特征抽取和识别实验表明,该特征抽取方法具有尺度和旋转不变性,有较好的抗干扰性和很好的分类性能,实用价值很高。
In pressed character recognition,the character is a protuberant character on the difference of reflectance.So,the feature extraction method based on the binary-scale character can't adapt for the protuberant character.To solve this problem,a new method of direct gray-scale feature extraction for the protuberant character recognition is presented.First-ly,the gray-scale character image is transformed into a function of one independent variable in the circular projection space.Secondly,the derived one-dimensional pattern is decomposed a set of wavelet transformation sub-patterns.Finally,these sub-patterns are readily computed to use DCT.The new method keeps integrity feature of protuberant character information dramatically,and overcomes the main shortages of the traditional method on binary-scale image,such as to depend on a binary algorithm extremely.A limited set of the tag pressed protuberant characters is recognized by the new method.The results show that the proposed method is proved to be in scale and orientation invariant ,and can yield an excellent classification rate on the condition of noise and deformity.It has highly worth applying to OCR fields.
出处
《计算机工程与应用》
CSCD
北大核心
2004年第6期23-26,46,共5页
Computer Engineering and Applications
基金
山东省重点产业化项目(编号:0203c06)的资助
关键词
特征抽取
小波变换
DCT变换
圆周投影
凹凸字符识别
Feature extraction,wavelet transform,discrete cosine transform(DCT),circular projection,protuberant character recognition
