无约束手掌图像采集系统及相应特征定位算法  

Non-constraint Palm Image Acquisition System and Corresponding Feature Locating Algorithm

作　　者：戴雷[1] 

机构地区：[1]中国科学院半导体研究所,北京100083

出　　处：《数据采集与处理》2010年第2期183-187,共5页Journal of Data Acquisition and Processing

基　　金：国家"八六三"高技术研究发展计划(2006AA01Z123)资助项目;国家自然科学基金(60753001)资助项目

摘　　要：为解决高自由度手掌图像的采集和手掌特征定位问题,本文提出了新型的无约束手掌图像采集系统及相应的特征定位算法。采集系统以弧形采集区域对手掌进行采集,过程中对用户左右手类别,手掌放置的位置、角度,手掌形状和大小无限制。手掌图像特征定位算法利用肤色信息分割手掌区域,以修正手掌重心点为原点建立极坐标系,通过计算掌形边缘点在极坐标下的分布特征,得到指尖指凹点坐标和左右手判别信息,作为手掌特征定位的输出结果。实验表明,本系统所采集的手掌图像清晰,特征定位算法的正确率达到98.5%。A novel non-costraint palm image acquisition system is fabricated to obtain palm images with the high freedom degree and a corresponding feature locating algorithm is proposed. The collection area of the system is cambered to fit the hand shape. The system is independent for the type, the position, the angle, shape, and the gesture of the palm. The feature locating algorithm uses the skin information to obtain the palm area, and then the distribution characteristics of palm-edge points are calculated in polar coordinates. The outputs of the algorithm are the judgment of the palm type and the point locations of the finger tip and the hollow. Experimental results show that images obtained by the system have high quality, and the accuracy rate of the feature locating algorithm achieves 98. 5 %.

关 键 词：边缘提取 形态学 极坐标 修正重心 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]