摘要
手部特征点识别是手部尺寸测量的关键问题,本文给出了基于彩色图像的手部特征点自动识别方法。首先,给出了提取手部轮廓的算法,基本原理是利用人体皮肤的色调特征将手部彩色图像二值化并提取手部轮廓,其优点是图像二值化过程在YUV空间进行,不受测量个体肤色差异、光照、遮挡等因素的影响。然后,给出了在手部轮廓线上自动识别全部25个特征点方法,针对特征点的不同特点给出了不同的处理方法,位于指尖以及指缝末端处的特征点,使用了DOS方法进行自动识别,其他特征点则利用统计得到的经验公式进行计算。实验结果说明,本文给出的算法在手部特征点的自动识别率和准确率上都已经能够满足手部尺寸测量的需要,并且具有较强的鲁棒性,已成功应用于未成人和军人等的手部尺寸测量系统中。
Hand feature points identification is a key problem of hand size measurement. This paper provides a novel method to automatically identify hand feature points. First, an algorithm of hand contour extraction is presented in this paper, which uses the skin hue to binarize the hand color image and extracts the hand contour in YUV color space. The advantage of this approach is that the image binarization process will not be affected by individual differences of hand color, ill-lumination and partial occlusion of light. Then, this paper provides an approach to automatically identify all 25 hand feature points from the hand contour. This paper gives two types of processing methods to identify different feature points: (1) The feature points located on finger tips and finger webs are identified with the DOS method; (2) other feature points are computed with statistical formulas. The results show that the algorithm provided in this paper could not only meet the requirements of hand size measurement system both in recognition rate and in precision, but also possess strong robustness. It has been applied successfully in the hand size measurement systems of non-adults and soldiers.
作者
董娜
DONG Na(CETC 15 Institute, Beijing 100083, China)
出处
《软件》
2017年第3期97-103,共7页
Software
关键词
手部测量
轮廓提取
特征识别
手部特征点
Hand measurement
Contour extraction
Feature recognition
Hand feature points
