指纹图像的定量化区域矢量编码与识别

Regional geometric feature coding of fingerprint

为了识别一幅图像 ,通常需要对它进行编码 ,提取特征信息。大体上可以归纳为两类方法 :统计方法和语言(或结构 )学方法。两类方法各有优缺点。第一类方法很少利用图像本身的结构关系 ,而第二类方法则没有考虑图像在环境中所受的噪声干扰。因此把二者结合起来 ,各取其长 ,是可取的途径。在统计方法的基础上 ,考虑了指纹的结构特点 ,按纹理方向进行分组 ,对它们实行定量化编码 ,并由此建立了方向矢量样本形式 ,给出了试验结果。这种方法的样本数据量小 ,构造过程简单 ,匹配和识别的信噪比高 ,不易产生错误匹配和判别。

Image coding is a method to identify subject It generally includes the statistics and structure method In order to enhance the precision,this paper researched the quantified coding Based on the former two methods,a regional geometric feature coding is developed in the paper According to the structure of fingerprint,the geometric feature samples are built and the matching results are given This coding method includes fewer datas The precision is high It can be used in many kinds of image recognition The rapid and exact identification system is cried for in many fields This method will be applied broadly in future