计算机应用研究2017，Vol.34Issue(1)：302-305,320,5.DOI:10.3969/j.issn.1001-3695.2017.01.068
基于时空图像分割和交互区域检测的人体动作识别方法
Human action recognition method based on spatio-temporal image segmentation and interactive area detection
摘要
Abstract
For the issues that the existing human action recognition method has not considered the role of non human targets, this paper proposed a human action recognition scheme base on spatio-temporal image segmentation and interactive area detec-tion.First,this scheme detected the human body contour in the video stream,and formed the key section by spatio-temporal image segmentation.Then,it extended segment to make it contain a non human object interaction with human body.After that, it used spatio-temporal histogram of oriented gradient(HOG)and histograms of optical flow(HOF)descriptors to represent the static and dynamic characteristics of the key segments,and used K-means algorithm to build a code book,while it used local linear encoding(LLC)technology to optimize the code book.Finally,it used the nonlinear support vector machine (SVM)to study the characteristics of the feature and perform the action recognition.Experimental results show that the proposed scheme has a higher recognition rate compared with the existing methods.关键词
人体动作识别/时空图像分割/交互区域/局部约束线性编码/支持向量机Key words
human action recognition/spatio-temporal image segmentation/interactive area/locality-constrained linear coding/support vector machine分类
信息技术与安全科学引用本文复制引用
张杰,吴剑章,汤嘉立,范洪辉..基于时空图像分割和交互区域检测的人体动作识别方法[J].计算机应用研究,2017,34(1):302-305,320,5.基金项目
国家自然科学基金资助项目（61302124）；江苏省自然科学基金资助项目（BK20130235）；江苏省产学研联合创新基金资助项目（BY2014038-05）；江苏省高校自然科学研究面上项目 （）
