一种快速的手掌轮廓特征点提取方法  

A Rapid Feature Points Extraction Method for Palm Outline

作　　者：金璟璇[1] 王红纪[2] 崔荣一[1] 

机构地区：[1]延边大学计算机科学与技术系智能信息处理研究室,延吉133002 [2]漯河职业技术学院计算机工程系,漯河462002

出　　处：《武汉理工大学学报》2010年第23期154-156,178,共4页Journal of Wuhan University of Technology

基　　金：吉林省科技厅资助项目(20050703-1)

摘　　要：快速而准确地提取手掌轮廓特征点是精确提取在线掌纹图像感兴趣区域的关键。为有效地解决这一问题,提出了快速的手掌轮廓特征点的提取方法。首先,把掌纹图像的灰度直方图谷底对应的灰度值作为全局阈值将图像二值化;其次,利用Log算子检测出手掌的轮廓线并进行跟踪以获得轮廓线的8-邻域链码;最后,根据8-邻域链码在特征点附近的变化规律提取出手掌轮廓特征点。为了验证该方法的有效性,根据提取出的特征点对300幅掌纹图像的感兴趣区域按比例进行了定位分割。实验结果表明该方法效率高且具有良好的平移、旋转不变性。Extracting feature points quickly and accurately is the key to extract ROI（region-of-interest） from palmprint image accurately.In order to solve this problem,a rapid method to extract feature points of palm outline was proposed. Firstly,palmprint image was converted into binary image,the global threshold of binary process was the gray value corresponding to the valley bottom of gray histogram.Then,Log operator was used to detect palm outline,subsequently 8neighborhood boundary-tracking algorithm was used to obtain palm outline chaincode.Finally,feature points of palm outline was extracted according to the variational regularity of chain-code.To demonstrate the effectiveness of the proposed method,ROI was extracted in proportion from 300 palmprint image according to the extracted feature points.The experiment results demonstrate that the proposed method is high-efficency and possesses translation and rotation invariant property.

关 键 词：掌纹图像 分割 边界链码 特征点提取 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]