公开(公告)号：KR1020020031724A

公开(公告)日：2002-05-03

申请号：KR1020000062398

申请日：2000-10-23

Applicant: 한국전자통신연구원

Inventor： 임신영 ,  이관용 ,  조성원

IPC: G06K9/46

CPC classification number: G06K9/00617 ,  G06K9/0061 ,  G06K9/4614 ,  G06K9/6267

Abstract: PURPOSE: A characteristic extraction and recognition device for an iris pattern process and a method thereof are provided to effectively extract the characteristic from an iris image by using the multiple division of wavelet conversion and to utilize the improved LVQ(Learning Vector Quantization) applying the weight initialization and winner choice mode. CONSTITUTION: The iris characteristic extraction and recognition device comprises an iris input tool(110) receiving the input of iris image signal, a characteristic extraction tool(120) performing the multiple division by wavelet converting the iris image signal inputted from the input tool and extracting a characteristic vector by quantizing the characteristic vector information, an LVQ recognition tool(130) executing the recognition through the approved LVQ applying the weight initialization and winner choice mode by receiving the characteristic vector from the characteristic extraction tool, and an identification confirmation tool(140) confirming the identification based on the recognition result.

Abstract translation: 目的：提供一种用于虹膜图案处理的特征提取和识别装置及其方法，以通过使用小波变换的多重分割来有效地从虹膜图像提取特征，并利用应用权重的改进的LVQ（学习矢量量化） 初始化和赢家选择模式。 构成：虹膜特征提取识别装置包括接收虹膜图像信号的输入的光圈输入工具（110），通过小波变换进行多重分割的特征提取工具（120），该输入工具从输入工具输入的虹膜图像信号和提取 通过量化特征向量信息的特征向量，LVQ识别工具（130），通过从特征提取工具接收特征向量，通过应用权重初始化和胜者选择模式的批准的LVQ执行识别，以及识别确认工具（140 ）基于识别结果确认识别。