CAJ | 학술논문

在连续手势自动识别中,如果可以借助自然语言对手势进行描述,这将更接近于人们对手势认知的方法,这在一定程度上将有助于提高识别的准确性。通过使用模糊集与条件随机场相结合的方法实现了对连续手势的标注、分割。 FCRF首先采用模糊集对手势进行描述,实现了自然语言对手势的描述,然后通过对条件随机场中状态和状态转移参数的修改实现了采用模糊特征的连续手势序列标注。实验结果显示,与现有的分割模型相比较,FCRF模型对手势序列标注具有较高的正确率,通过ROC特性曲线的分析,FCRF有较好的分类性能。
재련속수세자동식별중,여과가이차조자연어언대수세진행묘술,저장경접근우인문대수세인지적방법,저재일정정도상장유조우제고식별적준학성。통과사용모호집여조건수궤장상결합적방법실현료대련속수세적표주、분할。 FCRF수선채용모호집대수세진행묘술,실현료자연어언대수세적묘술,연후통과대조건수궤장중상태화상태전이삼수적수개실현료채용모호특정적련속수세서렬표주。실험결과현시,여현유적분할모형상비교,FCRF모형대수세서렬표주구유교고적정학솔,통과ROC특성곡선적분석,FCRF유교호적분류성능。
An original method for hand gesture labeling in automatic gesture labelling and segment using a fuzzyfied Conditional Random Fields ( FCRFs) were presented. Through combining CRFs and fuzzy set, model has not only all the advantages of CRFs in sequence labeling compared with Hidden Markov Model ( HMM) , but also the advantage of retaining the imprecise character of gestures. The efficiency of the proposed method was tested with unsegmented gesture sequences in three different hand gesture data sets. The experiment results demonstrate that this fuzzy-based conditional random fields compares favorably to some other model on hand gesture recognition tasks.