摘要: 在传统的计算机视觉领域中,底层任务被认为是自主的、自底向上的过程,造成较低的图像识别率。为此,提出一种基于拓扑知觉理论的人脸表情识别方法。该方法把人脸具有拓扑不变性的性质用于人脸拓扑轮廓的提取,将提取的特征与主成分分析相结合,作为人脸大范围特征信息,将大范围优先原理应用于人脸表情的识别算法中,设计RBF+Adaboost多层分类器。实验结果表明,该方法可以提高人脸表情的识别率。
关键词:
人脸表情识别,
拓扑优先,
人脸拓扑轮廓,
大范围优先,
自适应增强
Abstract: On traditional computer visual field, the task is widely considered to be independent bottom-up, this causes low recognition rate of image. This paper proposes the facial expression recognition method based on the topology consciousness theory. The method applies the stability of human face topology invariance to abstract the facial outline. And adds the PCA to integrate as the facial large extent characterized information, applies large range priority principle to facial expression recognition, and designs the RBF+Adaboost classification. Experimental results show this method can improve the rate of facial expression recognition.
Key words:
facial expression recognition,
topological priority,
face topological contour,
large range priority,
adaptive enhancement
中图分类号:
王晓峰, 张丽君. 基于拓扑知觉理论的人脸表情识别方法[J]. 计算机工程, 2012, 38(06): 193-195.
WANG Xiao-Feng, ZHANG Li-Jun. Facial Expression Recognition Method Based on Topological Perception Theory[J]. Computer Engineering, 2012, 38(06): 193-195.