摘要
以检测不良坐姿,分析人们学习工作状态为引导,设计了基于深度图像的坐姿检测系统;该系统采用3D传感器获取人体坐立时的深度图像,设计了一种基于深度阈值的快速前景提取与干扰移除方法,能快速有效的提取坐姿状态下的人体分割图;基于人体轮廓的曲线特征实现了人体关键点定位,通过人体关键点的角度、深度信息与轮廓特征,对不同的坐姿图像进行统计分析,得到了一种区分不同坐姿的判定基准,运用该基准对9种不同坐姿进行识别,平均识别率可达到90%;最后,基于Android平台设计了坐姿检测系统的应用软件,实现了坐姿检测、坐姿提醒以及姿态统计等功能,并且对学习过程进行测试,测试结果表明,本系统可以有效的检测出9种坐姿、并对不良坐姿进行提醒与统计。
For the purpose of the detection of bad postures and analysis of people＇ s learning and working states, a sitting detection sys tern based on depth images is designed. The system uses the 3D sensor to obtain the depth image of the sitting human body, the rapid fore ground extraction and interference removal method based on the depth threshold is designed, which can quickly and effectively extract the hu man body segmentation map in the sitting state. Based on the characteristics of the contours of the human body, the key points o[ the human body are located. Based on the angles of human body key points, depth information and contour features, the different sitting images are sin tistically analyzed, and a criterion for distinguishing the different sitting positions is obtained. 9 different sitting positions are identify, the av erage recognition rate can reach 90%. Finally, an Android platform-based sitting position detection system application is designed to realize functions including sitting position detection, posture reminder and attitude statistics and other functions. The APP is used to test the learn ing process, test results show that the system can effectively detect 9 kinds of sitting postures, make reminds to bad posture and do statistics at well.
出处
《计算机测量与控制》
2017年第9期8-12,共5页
Computer Measurement &Control
作者简介
曾星（1993-），男，四川遂宁人，硕士研究生，主要从事传感器与战场环境监测方向的研究。
罗武胜（1972-），男，湖南祁东人，教授，博士研究生导师，主要从事传感器与战场环境监测方向的研究。
