基于手指骨架提取的手形识别方法  

A method of hand-shape recognition based on extraction of finger skeleton

作　　者：李温温[1,2] 刘富[1] 高雷[1] 

机构地区：[1]吉林大学通信工程学院,吉林长春130022 [2]白城师范学院机械工程学院,吉林白城137000

出　　处：《中南大学学报（自然科学版）》2016年第3期777-783,共7页Journal of Central South University:Science and Technology

基　　金：国家自然科学基金资助项目(51105170);吉林省科技发展计划项目(SC0701028)~~

摘　　要：针对使用非接触式手形采集仪采集手形时,手掌图像的指根处轮廓会因手掌张开幅度不同而出现不同程度的变形,增加手形精确定位和手形识别的难度这一问题,提出一种不受指根处轮廓变形影响的手形识别算法。该算法首先提取出细化的手指骨架,其次拟合手指骨架并定位手指中线,然后提取与指根部无关的指形特征,最后,计算注册样本与待识别样本的差异特征统计值并进行判别决策。研究结果表明:本文提出的手形识别算法对指根处轮廓变形有着较强的鲁棒性,识别率可达99.62%。When a non-contact hand shape acquisition instrument is used to collect hand shape, there will be a kind of deformation at different levels in the finger root contour because of the difference in open degree of palm, which increases the difficulty of the accurate positioning and recognition hand-shape. In order to solve this problem, a new method for hand-shape recognition was presented, which was not affected by the deformation of the finger root contour. Firstly, the thinning finger skeleton was extracted, and then the finger central axis and the binomial fitting finger skeleton were located. The finger geometric features unrelated to the finger root contour were extracted. Finally, the statistics of differences characteristic between the registered samples and identifying samples were calculated to recognize the hand shape. The results show that the proposed method is robust to the deformation of the finger root contour, and the recognition rate can reach 99.62%.

关 键 词：手形识别 手指细化 骨架拟合 手指几何特征 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]