基于谱特征提取算法的海员表情识别方法研究

Seamen’s expression recognition based on spectrumfeature extraction algorithm

基于及时了解和调整海员的工作状态,避免海上交通事故频繁发生的目的,采用了基于空间一致性的谱特征提取表情识别算法。谱特征提取算法是在谱聚类算法理论基础上,通过在进行谱特征提取之前加空间一致性进行约束来改进实现的。通过对海员特定表情(比如疲劳、焦虑等)的识别试验,得出该算法能够有效地提取表情的特征信息,特别是对疲劳焦虑等负面情绪的表情,能有效提高识别率。

In order to understand and adjust the seamen＇s working state,avoid maritime traffic accidents occur frequently, adopt facial expression recognition based on the spatial coherence spectrum feature extraction algorithm. On the basis of spectral clustering algorithm, spectrum feature extraction algorithm is improved by space consistency constraint before spectrum feature extraction. By the identification of the seamen＇s specific expression （such as fatigue,anxiety,etc.）, it show that, the algorithm can effectively extract characteristic information of expression, especially for the negative emotions of fatigue and anxiety, and effectively improve the recognition＇s rate.