一种便于自动生成识别规则的图象基元抽取方法

A FEATURE EXTRACTION METHOD FOR EASILY PRODUCING IMAGE RECOGNITION RULES

  • 摘要: 本文提出了一种便于自动生成识别规则的图象基元抽取方法.并将其用于手写字符的识别.结果表明,用本方法得到的图象基元链(句子)对图象的位置、大小、局部形变不敏感,降低了语言的信息冗余度.因而易于由样本图象直接生成识别规则.

     

    Abstract: For feature extraction of an image to be recognized this paper presents a newmethod by means of which recognition rules can be produced autcmatically.An experiment on hand-character recognition with this method has been carriedout.The result is quite satisfactory.It has been shown that the feature-chains(sen-tences)extracted by this method is not sensitive to position,size and deformationof the images to be recognized,And the redundant information of languages decreases.So it is easy to produce recognition rules from image samples.

     

/

返回文章
返回