计算机工程与应用 ›› 2015, Vol. 51 ›› Issue (13): 197-201.

• 图形图像处理 • 上一篇    下一篇

移动终端手势识别中DTW匹配算法研究

王杰锋,周治平,苗敏敏   

  1. 江南大学 物联网工程学院,江苏 无锡 214122
  • 出版日期:2015-07-01 发布日期:2015-06-30

Study of DTW matching algorithm on mobile terminal gesture recognition

WANG Jiefeng, ZHOU Zhiping, MIAO Minmin   

  1. School of Internet of Things Engineering, Jiangnan University, Wuxi, Jiangsu 214122, China
  • Online:2015-07-01 Published:2015-06-30

摘要: 针对现有手势识别算法中存在的计算量复杂、识别率低和效率差等不足,提出一种基于L-ADTW的移动终端手势识别的匹配方法,利用手机内置加速度传感器进行手势识别,采用差分检测方法自动检测出有效手势序列,用改进的起始松弛点对有效手势序列进行校准,使用L-ADTW匹配算法进行有效手势序列段匹配,并将匹配结果与优化阈值比较分析。实验结果表明,提出的L-ADTW算法,在保证手势匹配高效的前提下,使等错率降到2.5%,认证成功率提高了5.37%。

关键词: 生物特征, 手势识别, 加速度传感器, 差分检测, 基于改进的起始松弛点和绝对值距离的动态时间规整算法(L-ADTW)

Abstract: For gesture recognition algorithm has complex calculation, poor recognition rate and efficiency rate, this paper proposes a mobile terminal gesture recognition matching method based on L-ADTW, using the phone in-built accelerometer to carry out gesture recognition, adopting differential detection effective method to automatically detect the gesture sequence, employing the improved starting lax point for correcting gesture sequence. Then it uses the L-ADTW matching algorithm for effective gesture sequence to match, and analyzes the results of the matches compared with the optimal threshold. Experiments show that the proposed L-ADTW algorithm, under the premise of gesture matching efficience, drops the equal error rate to 2.5%, and the authentication success rate increases 5.37%.

Key words: biometric features, gesture recognition, acceleration sensor, differential detection, Lax points and Absolute Dynamic Time Wrap(L-ADTW)