现代电子技术2016，Vol.39Issue(15)：38-42,46,6.DOI:10.16652/j.issn.1004-373x.2016.15.010
复杂场景下面向时空模糊性的人体动作检测方案研究
Scheme of human motion detection oriented to space-time fuzziness in complex scene
摘要
Abstract
Unlike the traditional human motion detection in well⁃controlled environment,the space and time boundary exists the space⁃time fuzziness due to the background noise,human body occlusion and incomplete tracking while performing mo⁃tion detection in complex scene. The available motion detection schemes can′t solve the above problem effectively,therefore the motion history image(MHI)features and appearance features are used to distinguish the human motion. And then the candidate regions of an action are regarded as an instance package,and the simulated annealing multiple instances learning support vector machines (SMILE⁃SVM) algorithm is proposed for realizing the human motion detection. The simulation results show that the proposed algorithm is superior to the available algorithms in the aspect of public CMU action dataset. In addition,a client inten⁃tion detection system for supermarkets is proposed,which can detect whether the customers intend to get the merchandise on shelf in crowded supermarket,and has the significant value for merchants to research the customer interests.关键词
人体动作检测/时空模糊性/运动历史图像特征/外观特征/多实例学习Key words
human motion detection/space-time fuzziness/motion history image feature/appearance feature/multi-instance learning分类
信息技术与安全科学引用本文复制引用
从继成..复杂场景下面向时空模糊性的人体动作检测方案研究[J].现代电子技术,2016,39(15):38-42,46,6.基金项目
国家自然科学基金 （）
