摘要
针对多组手语语句中重复出现的手语单词识别问题,提出了一种识别方法。该方法利用时间规整算法构建手语识别模型,并通过条件迭代算法快速计算最大后验概率。在南佛罗里达大学公共手语数据集进行实验,证明了该方法具有一定的实用性。
For the problem of sign language recognition in continuous sentences, a method was proposed. The sign language recognition model was established with Dynamic Time Warping （DTW）, and through the Iterated Conditional Modes（ICM）computed the maximum a posteriori probability. The performance of this method was assessed by computer simulations.
出处
《微型机与应用》
2015年第2期49-51,58,共4页
Microcomputer & Its Applications
关键词
手语识别
动态时间规整算法
条件迭代算法
sign language recognition
dynamic time warping
iterated conditional modes
