基于傅里叶-隐马尔科夫模型的人体行为识别方法研究

摘要

Human motion detection and behavior recognition have a wide range of applications,including artificial intelligence,computer vision and pattern recognition.Human behavior recognition has important application value in medical,commercial and military fields.In order to explore a good method of human behavior recognition,In this paper,we need to know the outline of the binary image of human behavior in the process of recognizing human behavior sequence images,and then take the Fourier transform in a scientific way and then transform into vector Observe the symbol sequence,the vector quantization to the eigenvector change,easy to extract the characteristics of the human contour,the subsequent application of research.Finally,the behavior of the human body to identify,using HMM classifier.The use of Fourier-hidden Markov model for human identification can effectively improve the recognition rate of human behavior.The average recognition rate of single test in this test reaches 94％.It is necessary to carry out in-depth exploration to identify complex movements in complex environments,Improve related work.%运动人体检测和行为识别涉及广泛,包括人工智能、计算机视觉、模式识别等,人体行为识别在医疗、商业、军事中具有重要的应用价值,为探究良好的人体行为识别方法,本文引入傅里叶-隐马尔可夫模型进行相关分析,在人体行为序列图像的识别过程中,需要了解有关人体行为二值图像的轮廓,然后采取科学的方式进行傅里叶变换,接着进行向量转化,形成观察符号序列,将矢量量化向特征向量变化,便于提取人体轮廓的特征,进行后续的应用研究.最后对人体的行为进行识别,采用隐马尔大夫分类器.利用傅里叶-隐马尔科夫模型进行人体识别,能够有效提高人体行为识别率,本次测试单个行为的识别中平均识别率达到94％,要进行深入探究,进行复杂环境复杂动作的识别,促进相关工作的改进.