摘要
针对传统人体运动检测方法存在的不足,提出一种图像深度和人体信息熵相结合的检测方法.首先依据图像的颜色和深度信息,结合Hough森林法提取人体部位的关键点;再利用关键点的信息连接出人体骨架,采用改进的信息熵表达式计算人体运动的熵值,并对数据做相关分析;最后,得到人体的运动状态.实验结果表明:采用的深度图像能更有效地提取人体关键点,对优化传统人体识别方法存在重要意义;所提出的信息熵法应用到人体运动检测中可以更加直观反映人体运动情况,明显提高检测效率和准确率.
Deficiencies against the traditional method of human motion detection, a combination of image depth and body information entropy of detection method is proposed. Firstly, information entropy method used the image of color and depth information and Hough forest to extract key points of human body parts. Secondly, The human skeleton can be gotten by the information of key points, and the entropy of human motion can be evaluated according to improved information entropy expression. Thereafter, the data can be analyzed. Finally, human motion state can be obtained. The results show that the key points are effectively extracted by depth pictures, which have great significance to optimize the traditional method of human recognition, and the information entropy method is applied to human movement detection can be more intuitive to reflect human motion, which significantly improve detection efficiency and accuracy.
出处
《小型微型计算机系统》
CSCD
北大核心
2014年第2期388-392,共5页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(61070024)资助
国家住建部科技项目(2010-K9-22)资助
关键词
信息熵
人体运动
深度信息
关键点
information entropy
human motion
the depth of information
key points
