关键词: 精神疲劳识别,面部特征,矛盾问题可拓模型,可拓策略生成,可拓方法

Abstract: There are contradictions between the non-intrusive,real-time requirements and recognition rate in facial feature-based mental fatigue recognition algorithms which hinder the practical application of the algorithms.To solve these problems,extension theories and methods were introduced to build an extension model of the contradiction problem.A dependent function used to measure contradictions degree of the problem and a function for strategies evaluation were proposed.Extension analysis and extension transformation were used to generate solving strategies for mental fatigue recognition.Research results show that,based on models and strategies raised in this paper,we can develop a variety of feature extraction and mental fatigue recognition algorithms that both meet the non-intrusive,real-time requirements and recognition rate.At the same time,the value of the strategy can be quantified and compared.Results of this study can improve the intelligence of mental fatigue recognition method,and provides a good example to solve the widespread conflicts between computational complexity and accuracy existing in pattern recognition algorithms.

Key words: Mental fatigue recognition,Facial features,Extension model for contradiction problems,Extension strategies generating,Extension methods