手掌图的像采集系统特征定位算法.pdf

252 Vol. 25 No . 2 2010 3 Journal of Data Acquisition Processing Mar . 2010 : 1004-9037(2010) 02-0183-05 戴　雷 ( , , 100083) : 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 统及相 应的特征定位算法。采集 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 , 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 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% 。 : 边缘提取; 形态学; 极坐标; 修正重心 :T P391.4 1: A - Non constraint Palm Image Acquisition System and Corresponding Feature Locating Algorithm D a i L e i ( Inst itute of emiconductors, Chinese Academy of ciences, Beijing, 100083, China) Abstract: A novel non-costraint palm image acquisition system is fabricated to obtain palm images w ith the high freedom degree and a corresponding feature locating algorithm is proposed T he 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%. Key words: edge abstraction; morphology ; polar coordinates; revisory gravity center