Physical education teaching auxiliary system based on motion identification technology and implementation method of physical education teaching auxiliary system

Info

Publication number
CN102243687A
Authority
CN
Grant status
Application
Patent type
Prior art keywords
education teaching
physical education
teaching auxiliary
teaching
education
Prior art date
Application number
CN 201110102571
Other languages
Chinese (zh)
Inventor
刘思杨
宁岩
牛涛
王略志
陈拥权
Original Assignee
安徽寰智信息科技股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date

Links

Abstract

The invention provides a physical education teaching auxiliary system based on a motion identification technology and an implementation method of the physical education teaching auxiliary system. The system is applicable to a personal computer (PC) or an embedded host. The system comprises a movement data collection module, a movement data acquisition module, an identification and training module and a virtual teaching environment module. The device is characterized in that: a micro-inertia measurement unit and an inertia parameter extraction unit are arranged in the movement data collection module; and a movement information resolving unit transmits data which is output by the inertia parameter extraction unit to a multi-sensor data fusion unit for resolving. The movement situation of a target is reflected comprehensively in an inertia tracking mode and an optical tracking mode, so a tracking range is effectively expanded, measurement accuracy is improved, and the problems that the integral information of the target cannot be acquired in the inertia tracking mode, complicated movement identification cannot be performed and sensitivity is poor are solved. A brand new teaching mode is provided for physical education teaching, and a physical education teaching method is digitalized, multi-media and scientifically standardized.

Description

一种基于动作识别技术的体育教学辅助系统及其实现方法 The action taken PE teaching aids identification technology system implementation method based on its

技术领域 TECHNICAL FIELD

[0001] 本发明涉及人机交互、动作识别以及计算机辅助教学领域,尤其涉及一种体育教学辅助系统的构建方法。 [0001] The present invention relates to human-computer interaction, gesture recognition and computer-assisted instruction, and in particular relates to a method for constructing PE Teaching System.

背景技术 Background technique

[0002] 目前,运动追踪的方法根据传感元件的不同分为:惯性追踪、光学追踪、力和机械式追踪、电磁式追踪、声学追踪等等。 [0002] Currently, the motion tracking method based on different sensing elements are divided into: inertial tracking, optical tracking, force and mechanical tracking, electromagnetic tracking, acoustic track, and so on. 目前,在人机交互领域比较常用的运动追踪技术是惯性追踪技术和光学追踪技术。 Currently, in the field of human-computer interaction more commonly used motion tracking technology inertial tracking technology and optical tracking technology. 惯性追踪技术,通过在目标上设置惯性测量单元,测量得到加速度、角速度等数据,以此为基础使用数学工具解算,得到目标的运动情况。 Inertial tracking technology by providing an inertial measurement unit on the target, the measured acceleration, angular velocity and other data as a basis for solving mathematical tools give the movement goals. 惯性追踪的特点是实现简单,抗干扰性强;缺点是不能全面地获取被追踪目标的运动特征,只能有限和局部地反映被追踪目标的运动特点。 Inertia track is characterized by simple, strong anti-interference; the disadvantage is not fully access the tracked target motion feature, only limited and partial reflection of the tracked target motion characteristics. 光学追踪技术,通过对目标上特定光点的监视和跟踪来完成运动追踪的任务。 Optical tracking technology, through a specific spot on the target of surveillance and tracking motion tracking to complete tasks. 理论上,对于空间的任意一个点,只要它能同时被两台摄像机所见, 则根据同一瞬间两摄像机所拍摄的图像和摄像机参数,即可以确定这一时刻该点的空间位置。 Theoretically, for any space of a point, as long as it can be seen at the same time two cameras, and camera image according to the same instant two parameters captured by the camera, which can determine the spatial position of the point of this moment. 当摄像机以足够高的速率连续拍摄时,从图像序列中就可以得到该点的运动轨迹。 When the camcorder at a sufficiently high speed continuous shooting, the image sequence can get the trajectory of the point. 其中图像的处理采用的是三维图象重建技术,即通过摄像机记录图像,通过数字化处理形成虚拟物体,然后通过三维空间标定,确定物体的空间位置。 Wherein the image processing using the three-dimensional image reconstruction technique, i.e., an image recorded by the camera, the virtual object is formed by digital processing, and three-dimensional space by means of calibration, to determine the spatial location of the object. 光学追踪技术的特点是能全面地反映物体的运动情况,精度高;缺点是实现较为困难,且追踪的范围较小。 Characteristics of the optical tracking technology that can fully reflect the movement of objects, high precision; drawback is more difficult to achieve, and less tracking range.

[0003] CN10115888公开了一种基于计算机视觉的虚拟体育系统及其实现方法,用于通用计算机,利用计算机视觉来识别人体以及体育器械的运动状态和动作模式,并将动作模式反馈给计算机,通过计算机的处理,控制虚拟体育运动中的角色做出相应的动作。 [0003] CN10115888 discloses a computer vision based virtual sports system and its implementation method for a general purpose computer, using computer vision to identify the state of human motion and sports equipment and the operation mode and operation mode feedback to the computer through computer processing, control virtual sports roles make the appropriate action. 该发明可以更好地提高全民健身运动的普及程度,但缺点是对于实际动作的识别范围较小。 The invention may be better to increase the popularity of national fitness campaign, but the drawback is the smaller the actual operation of the recognition range.

发明内容 SUMMARY

[0004] 本发明的目的是提供一种基于动作识别技术的体育教学辅助系统及其实现方法, 利用人机交互技术以及光学追踪与惯性追踪相结合的方法,能够有效扩大运动追踪范围并及时反馈运动信息,实现较大范围内特定运动信息的采集与处理,并将之应用于计算机虚拟辅助体育教学。 [0004] The object of the present invention is to provide a motion-based recognition technology in physical education support system and its implementation method, the use of human-computer interaction techniques and methods of optical tracking and tracing combination of inertia, can effectively expand the range of motion tracking and timely feedback motion information acquisition and processing to achieve a wide range of sport-specific information, and applied to a virtual computer Athletics Teaching.

[0005] 本发明解决其技术问题所采用的技术方案是:一种基于动作识别技术的体育教学辅助系统,用于PC或嵌入式主机,其包括运动数据采集模块、运动数据获取模块、辨识与训练模块、虚拟教学环境模块组成, [0005] Technical Solution The present invention solves the technical problem is adopted: based on motion recognition technology physical education support system for PC or embedded host, which includes motion data acquisition module, motion data acquisition module, and identification training module, virtual learning environment modules,

[0006] 所述的运动数据采集模块进一步包括:特定波长点光源、不少于二个摄像头; Motion data acquisition module [0006] wherein further comprising: a point light source of a specific wavelength, at least two cameras;

[0007] 所述的运动数据获取模块进一步包括:图像特征提取单元、三维空间定位单元、运动信息解算单元、多传感器数据融合单元; The movement of [0007] data acquisition module further comprises: image feature extraction unit, three-dimensional positioning unit, the motion information solver unit, multi-sensor data fusion unit;

[0008] 所述的辨识与训练模块进一步包括:动作模式数据采集模块、训练模块、辨识单元; Identification and Training Module [0008] The further comprises: an operation mode data acquisition module, training module identification unit;

[0009] 所述的虚拟教学环境模块包括一动作模式标准库、一动作解析单元、一虚拟教学环境、一显示设备; [0009] The virtual learning environment module comprises an operation mode of the standard library, a motion analysis unit, a virtual learning environment, a display device;

[0010] 其特征在于:在所述的运动数据采集模块中还是设有一微惯性测量单元和一惯性参数提取单元,微惯性测量单元与被测目标绑定,用于测量被测目标的惯性参数,供运动数据获取模块处理,所述的微惯性测量单元通过无线信号与所述的惯性参数提取单元信号输入端连接;所述的运动信息解算单元对惯性参数提取单元输出的数据传输到多传感器数据融合单元作解算处理。 [0010] wherein: said motion data acquisition module is provided with a micro inertial measurement unit and an inertial parameter extraction unit, micro inertial measurement unit and the measured binding targets for measuring inertial parameters measured target for processing motion data acquisition module, the micro inertial measurement unit is connected via a wireless signal and the inertial parameter extraction unit signal input terminal; the motion information solver unit data transmission inertia parameter extraction unit output to multiple sensor data fusion unit for solving process.

[0011] 所述的运动数据获取模块,其中所述的三维空间定位单元利用标记点在两个摄像头内成像的位置,利用双目视觉算法得到标记点的三维空间坐标,所述多传感器数据融合单元采用基于DS证据理论的多传感器数据融合算法对运动目标的惯性参数和三维空间坐标作处理,获取目标的动作模式数据。 [0011] The motion data acquisition module, wherein the three-dimensional positioning unit utilizes two markers in the camera imaging position, binocular vision algorithms to obtain three-dimensional coordinates of the markers, the multi-sensor data fusion unit based on multi-sensor data fusion DS evidence theory algorithm inertia parameters of moving objects and for the three-dimensional spatial coordinates, obtaining the goal of the operation mode data.

[0012] 所述的虚拟教学环境模块,其中: [0012] The virtual learning environment module, wherein:

[0013] 所述动作模式标准库用于供动作解析单元解析动作模式的范本;所述的动作模式标准库包括特定体育运动动作模式标准库、非特定动作模式标准库和连续动作模型库; [0013] The operation mode of the standard library for parsing unit parsing template for action operation mode; the operation mode of the standard library includes a specific sport operation mode of the standard library, non-specific mode of action of the standard library and continuous action model library;

[0014] 所述动作解析单元用于解析动作模式数据的辨识结果; [0014] The operation means for parsing the identification result of the parsing operation mode data;

[0015] 所述虚拟教学环境为人机互动平台,用于提供体育教学方案; [0015] The virtual learning environment for the man-machine interactive platform for providing sports education programs;

[0016] 所述显示设备用于反馈运动动作模式辨识的解析结果、显示人机交互动作状态, 其输入端与计算机连接; [0016] The analysis result display apparatus for feedback motion pattern recognition operation, the display operation state HCI, having an input connected to the computer;

[0017] 所述的特定体育运动动作模式标准库,以体育运动项目的基本动作为主,包括乒乓球的基本动作、网球基本动作、高尔夫挥杆动作、保龄球掷球动作、跑步的抬腿动作等; [0017] The specific mode of action sports standard library to basic operation of sports-based projects, including the basic movements of table tennis, tennis basic movements, golf swing, throw a bowling action, action running leg Wait;

[0018] 所述的非特定动作模式标准库,是根据体育教学要求定制特殊应用的动作模型库,包括鞍马运动员基本动作、体操、体育教学动漫游戏类应用的动作等; [0018] The non-specific mode of action of the standard library, based on PE teaching customized application-specific action model library, including the pommel horse athletes basic movements, gymnastics, physical education class animation and game applications and other actions;

[0019] 所述的连续动作模型库,为辅助体育教学特定功能而设计的由系列动作组成的组合动作库,它基于特定动作模型库和非特定模型库,包括体操、舞蹈、武术、健身气功等一套完整的动作。 Continuous action model library [0019], wherein the auxiliary sports teaching specific functions designed by the series of actions consisting of a combination of action library, which is based on specific and non-specific action model library model database, including gymnastics, dance, martial arts, Qigong a complete set of actions.

[0020] 一种基于动作识别技术的虚拟体育教学辅助系统的构建方法,其包括以下步骤: [0020] Based on the virtual PE teaching method action recognition technology to build support system, comprising the steps of:

[0021] A、将标记点与被测目标绑定,将摄像头安装在指定区域的正前方,用于捕捉指定区域的图像、获取标记点运动的惯性参数; [0021] A, the markers and the measured binding target, install the camera in front of the designated area for capturing an image of the designated area, obtain the inertia parameter markers Movement;

[0022] B、采集数字图像并输入通用计算机,得到目标运动的数字图像;采集标记点运动的惯性参数; [0022] B, digital image capture and enter a general purpose computer, a digital image of the target to give movement; inertia parameters acquisition marker movement;

[0023] C、对采集的惯性参数和数字图像进行解算和多传感器融合,获取标记点运动方式; [0023] C, inertia and digital image acquisition parameters were solver and multi-sensor fusion, get marker movement;

[0024] D、根据标记点运动方式识别其运动模式,传送给虚拟体育教学环境,并反馈输出相应的交互动作状态,对输出的交互动作状态进行统计,并与预设的体育教学指标进行比对和解析,得到动作修正方案如位置、速度、角度等方面,并将结果输出在显示设备上。 [0024] D, according to the motion markers identify its Sport mode, the transmission to the virtual PE teaching environment, interaction and feedback output corresponding operating state, the output of the interaction state statistics, and the preset physical education indicators than and resolve to give the correction action plan such as position, speed, angle, etc., and outputs the result on a display device.

[0025] 所述的步骤C还包括:在图像特征提取单元,首先通过二值算法提取目标运动的特征构图,对特征构图求取其重心,得到特征构图中心位置;接着根据对所选特征的计算, 建立特征之间的对应关系; Step C [0025] further comprises: image feature extraction means for extracting a feature patterned first target motion by binarizing method, wherein the composition for obtaining the center of gravity, wherein the composition obtained center position; then based on the selected features computing, establish correspondence between the features;

[0026] 所述步骤C还包括:对所述步骤B中获取的标记点运动方式和惯性参数,通过DS 证据理论的多传感器数据融合算法,得到反映目标运动特征的一致性数据。 [0026] said step C further comprising: markers motion and inertia parameters of the step (B) acquired by multi-sensor data fusion algorithm DS evidence theory to obtain consistent data reflect the characteristics of the target movement. [0027] 所述步骤D还包括: [0027] D further comprises the step of:

[0028] D1、采集各种样本动作模式数据,对采集到的样本动作模式数据进行标注; [0028] D1, collect a variety of data sample operation mode, the collected sample data tagging operation mode;

[0029] D2、逐一从所述样本动作模式数据中提取出反映其本质特征的特征向量; [0029] D2, each of which reflects the essential characteristics of the extracted feature vector from the sample data in the operation mode;

[0030] D3、根据所述特征向量划分所属类别区域,使得划分后的各个不同类别区域中只包含同类样本的特征向量,建立从特征向量到所属类别之间映射关系的分类器; [0030] D3, according to the feature vector divided Category region so that the various categories of zoning after the sample contains only similar feature vectors from the feature vector to establish mapping relationship between Category classification;

[0031] D4、对待检测的动作模式数据进行处理,提取其特征向量; [0031] D4, to treat motion pattern detection data is processed to extract the feature vector;

[0032] D5、将待检测的作模式数据的特征向量输入到所述分类器,分类器根据其特征向量进行判别,得到该待检测的作模式数据的辨识结果。 [0032] D5, will be detected as a feature vector pattern data is input to the classifier, classifier judgment is made based feature vectors to obtain the recognition result to be detected as the pattern data.

[0033] D6、动作模式数据的辨识结果输入到虚拟教学环境和动作解析单元中,通过显示设备,反馈输出运动状态。 [0033] D6, the operation mode of the recognition result data input to the virtual environment and the operation instruction parsing unit, through the display device, the output of the feedback motion.

[0034] 所述步骤D2还包括: [0034] the step D2 further comprises:

[0035] D21、对所述样本动作模式数据进行预处理,得到训练数据; [0035] D21, the operation mode of the sample data preprocessing to obtain training data;

[0036] D22、从训练数据中提取反映训练数据本质特征的特征成分; [0036] D22, extract the essential characteristics of the training data reflect the characteristics of components from the training data;

[0037] D23、将所述特征成分进行组合,得到所述特征向量。 [0037] D23, the characteristic components are combined to obtain the eigenvectors.

[0038] 所述步骤D4还包括: [0038] D4 further comprises the step of:

[0039] D41、对所述待检测的动作模式数据进行预处理,得到辨识数据; [0039] D41, for the operation to be detected pattern data preprocessing to obtain identification data;

[0040] D42、从辨识数据中提取反映辨识数据本质特征的特征成分; [0040] D42, extract data reflect the essential characteristics to identify the characteristics of components from the identification data;

[0041] D43、将所述特征成分进行组合,得到所述特征向量。 [0041] D43, the characteristic components are combined to obtain the eigenvectors.

[0042] 所述步骤D6还包括: [0042] D6 further comprising the step of:

[0043] D61、自定义不同类别的标准动作模式,建立动作模式标准库; [0043] D61, customize different classes of standard operation mode, the operation mode of the establishment of the standard library;

[0044] D62、将标准动作模式和体育运动教学步骤自定义为教学方案,并输入虚拟体育教学环境中; [0044] D62, the standard operation mode and sport teaching steps to customize the program for teaching and enter the virtual PE teaching environment;

[0045] D63、将辨识结果输入到虚拟体育教学环境中,实现人机交互运动状态;以动作模式标准库为参照标准,对辨识结果进行解析,得到动作修正方案; [0045] D63, the recognition result is input to the virtual PE teaching environment of interacting with the state of motion; in standard operation mode library as a reference standard analytical results of identification to obtain a correction action plan;

[0046] D64、将动作修正方案和动作模式数据辨识结果输入到显示设备。 [0046] D64, the correction action plan and action pattern data recognition result input to the display device.

[0047] 本发明的有益效果:由于采用在原来的光学追踪方式下增加了惯性追踪方式其综合后使反映目标运动情况,有效扩大了追踪范围,提高了测量的精度,解决了惯性追踪无法获取目标整体信息、不能做复杂的运动识别、敏感性差的问题,同时也解决光学追踪技术还原真实性差、有效追踪范围小以及阻挡影响的问题。 [0047] the beneficial effects of the invention: As in the original optical tracking method increases the inertia track approach to comprehensive post to make it reflect the objectives movement, effectively expanding the tracking range, improve the measurement accuracy, to solve the inertia track can not get overall objective information, can not do complex motion recognition, poor sensitivity of the issue, but also solve the optical tracking technology to restore the authenticity of the poor, small-scale and effective tracking of barrier impacts. 本发明还具有很强的实用性,为体育教学提供一种全新的教学模式,让体育运动教学方法趋于数字化、多媒体化和科学标准化。 The present invention is also highly practical, to provide a new teaching model of physical education, sports teaching methods tend to make the digital, multimedia and scientific standardization.

[0048] 以下将结合附图和实施例,对本发明进行较为详细的说明。 [0048] The accompanying drawings and the following embodiments, the present invention will be described in detail.

附图说明 BRIEF DESCRIPTION

[0049] 图1为本发明基于动作识别技术的体育辅助教学系统示意框图。 [0049] FIG. 1 is a schematic block diagram of the invention is based on PE teaching system assisted motion recognition technology.

[0050] 图2为本发明运动数据采集模块示意框图。 [0050] FIG. 2 is a schematic block diagram schematically motion data acquisition module.

[0051] 图3为本发明运动数据获取模块示意框图。 [0051] Figure 3 is a schematic of an athletic block diagram of a data acquisition module.

[0052] 图4为本发明辨识与训练模块示意框图。 [0052] FIG. 4 of the present invention and identification training module schematic diagram.

[0053] 图5为本发明虚拟教学环境模块示意框图。 [0053] FIG. 5 of the present invention virtual teaching environment block schematic diagram.

[0054] 图6为图5的局部放大示意框图。 [0054] FIG. 6 is a partial schematic block diagram of FIG. 5 enlarged. [0055] 图7为本发明总体结构的示意框图。 [0055] FIG. 7 is a schematic block diagram showing the overall structure of the invention. 具体实施方式 detailed description

[0056] 实施例1,如图1、图7所示,一种基于动作识别技术的体育教学辅助系统,用于PC 或嵌入式主机,其包括运动数据采集模块1、运动数据获取模块2、辨识与训练模块3、虚拟教学环境模块3组成; [0056] Example 1, as shown in Figure 1, shown in Figure 7, based on motion recognition technology physical education support system for PC or embedded host, including a motion data acquisition module 1, motion data acquisition module 2, identification and training module 3, virtual learning environment composed of 3 modules;

[0057] 如图2所示,所述的运动数据采集模块1进一步包括:特定波长点光源11、不少于二个摄像头12 ; [0057] 2, the athletic data collection module 1 further comprising: a specific wavelength of light points 11, 12 at least two cameras;

[0058] 如图3所示,所述的运动数据获取模块2进一步包括:图像特征提取单元21、三维空间定位单元22、运动信息解算单元23、多传感器数据融合单元M ; [0058] shown in Figure 3, the motion data acquisition module 2 further comprising: an image feature extraction unit 21, a three-dimensional positioning unit 22, the motion information solver unit 23, a multi-sensor data fusion unit M;

[0059] 运动数据获取模块2,其中所述的三维空间定位单元22利用标记点在两个摄像头12内成像的位置,利用双目视觉算法得到标记点的三维空间坐标,所述的多传感器数据融合单元M采用基于DS证据理论的多传感器数据融合算法对运动目标的惯性参数和三维空间坐标作处理,获取目标的动作模式数据。 [0059] motion data acquisition module 2, three-dimensional space in which the positioning unit 22 of the marker 12 in the two cameras within the imaging position, binocular vision algorithms to obtain three-dimensional coordinates of the markers, the multi-sensor data said fusion unit M based on multi-sensor data fusion DS evidence theory algorithm inertia parameters of moving objects and for the three-dimensional spatial coordinates, obtaining the goal of the operation mode data.

[0060] 如图4所示,所述的辨识与训练模块3进一步包括:动作模式数据采集模块31、训练模块32、辨识单元33 ; [0060] 4, said identification training module 3 further comprising: operation mode data acquisition module 31, training module 32, a recognition unit 33;

[0061] 如图5所示,所述的虚拟教学环境模块4包括一动作模式标准库41、一动作解析单元42、一虚拟教学环境43、一显示设备44 ; [0061] shown in Figure 5, the virtual learning environment module 4 includes a standard library operation mode 41, a motion analysis unit 42, a virtual learning environment 43, a display device 44;

[0062] 所述的虚拟教学环境模块4,其中:所述动作模式标准库41用于供动作解析单元解析动作模式的范本;所述动作解析单元42用于解析动作模式数据的辨识结果;所述虚拟教学环境43为人机互动平台,用于提供体育教学方案;所述显示44设备用于反馈运动动作模式辨识的解析结果、显示人机交互动作状态,其输入端与计算机连接; [0062] The virtual environment teaching module 4, wherein: said operation mode is the standard template library 41 for parsing unit for parsing the operation of the operation mode; parsing unit 42 for operation of the recognition result of the parsing operation mode data; the 43 said virtual learning environment for the human-machine interactive platform for providing sports education program; the display device 44 for motion analysis result of the feedback operation mode identification, display human-computer interaction operation state, having an input connected to the computer;

[0063] 如图2所示,在所述的运动数据采集模块1中还是设有一微惯性测量单13元和一惯性参数提取单元14,微惯性测量单元13与被测目标绑定,用于测量被测目标的惯性参数,供运动数据获取模块2处理,所述的微惯性测量单元13通过无线信号与所述的惯性参数提取单元14信号输入端连接;所述的运动信息解算单元23对惯性参数提取单元14输出的数据传输到多传感器数据融合单元M作解算处理。 [0063] FIG. 2, the motion data acquisition module 1 is provided with a micro inertial measurement unit 13 and $ 14 a inertia parameter extraction unit, micro inertial measurement unit 13 and the measured binding target for inertial measurement parameters measured object, motion data acquisition module for the second processing, signal input terminal 14 of the micro inertial measurement unit 13 by the inertial parameters of the wireless signal and the extraction unit is connected; the motion information computing unit 23 of the solution inertial parameter extraction unit 14 outputs the transmission data to the multi-sensor data fusion unit M for solving process.

[0064] 如图6所示,所述的动作模式标准库41包括特定体育运动动作模式标准库45、非特定动作模式标准库46和连续动作模型库47。 As shown in [0064] FIG. 6, the operation mode of the standard library includes 41 specific sports operation mode of the standard library 45, non-specific mode of action of the standard library 46 and continuous action model library 47.

[0065] 所述的特定体育运动动作模式标准库45,以体育运动项目的基本动作为主,包括乒乓球的基本动作、网球基本动作、高尔夫挥杆动作、保龄球掷球动作、跑步的抬腿动作等。 [0065] The specific mode of action sports standard library 45 to the basic operation of sports-based projects, including the basic movements of table tennis, tennis basic movements, golf swing, throw a bowling action, running leg action and so on.

[0066] 所述非特定动作模式标准库46,是根据体育教学要求定制特殊应用的动作模型库,包括鞍马运动员基本动作、体操、体育教学动漫游戏类应用的动作等. [0066] The non-specific mode of action of the standard library 46, based on PE teaching customized application-specific action model library, including the pommel horse athletes basic movements, gymnastics, physical education class animation and game applications and other actions.

[0067] 所述连续动作模型库47,为辅助体育教学特定功能而设计的由系列动作组成的组合动作库,它基于特定动作模型库和非特定模型库,包括体操、舞蹈、武术、健身气功等一套完整的动作。 [0067] The continuous action model library 47, auxiliary sports teaching specific functions designed by the series of actions consisting of a combination of action library, which is based on specific and non-specific action model library model database, including gymnastics, dance, martial arts, Qigong a complete set of actions.

[0068] 运动数据采集模块1主要用于采集运动目标的数字图像和惯性参数,该模块中特定波长点光源11为发出单色光的点光源,图像采集设备为可见光摄像头,微惯性测量单元13与被测目标绑定并通过无线信号与惯性参数提取单元14信号输入端连接,特定波长点光源11的光信号则由摄像头采集,摄像头12输出端与图像特征提取单元21输入端连接, 运动数据采集模块1采集的运动数据输入运动数据获取模块2中运动信息解析单元23和图像特征提取单元21,供运动数据获取模块2分析处理。 [0068] motion data acquisition module 1 is mainly used to collect digital images of moving objects and inertia parameters, the module specific wavelength point light source 11 to emit monochromatic light source, the image capture device is a visible light camera, Micro Inertial Measurement Unit 13 binding and measured target and extract the signal input unit 14 is connected via radio signals and inertial parameters specific point light source wavelength of the optical signal collected by the camera 11, camera 12 and the output of the image feature extraction unit 21 connected to the input terminal, the motion data motion data input motion data acquisition module 1 acquisition acquisition module 2 motion information analyzing unit 23 and the image feature extraction unit 21, motion data acquisition module 2 for analysis and processing.

[0069] 实施例2如图1、图7所示,一种基于动作识别技术的体育教学辅助系统,用于PC 或嵌入式主机,其包括运动数据采集模块1、运动数据获取模块2、辨识与训练模块3、虚拟教学环境模块4组成。 [0069] As shown in Example 1. Example 2, shown in Figure 7, based on PE Teaching System motion recognition technology for PC or embedded host, including a motion data acquisition module 1, motion data acquisition module 2, identification and 3 training modules, virtual learning environment composed of 4 modules.

[0070] 目标在所述显示设备44正面运动,与目标绑定的所述微惯性测量单元13将测量得到一组运动数据,经惯性参数提取单元14处理后得到运动惯性参数,通过无线传输模块传送至所述运动信息解算单元23 ;运动数据采集模块1中特定波长点光源11为发出单色光的点光源,图像采集设备为不少于二个的可见光摄像头,特定波长点光源的光信号则由摄像头采集,摄像头12输出端与图像特征提取单元21输入端连接。 [0070] In the target display device of the micro inertial measurement unit 44 positive movement, with the goal bound 13 measured a set of motion data, the inertia parameters treated inertial parameter extraction unit 14 after the wireless transmission module the motion information is transmitted to the solver unit 23; motion data acquisition module 1 in a specific wavelength point light source 11 to emit monochromatic light source, the image capture device not less than two of the visible light camera, light of a specific wavelength of light spot signal acquisition by the camera, the camera 12 and the output of the image feature extraction unit 21 connected to the input terminal. 运动数据采集模块1 采集的运动数据输入运动数据获取模块2中运动信息解析单元23和图像特征提取单元21, 供运动数据获取模块2分析处理。 Motion data input motion motion data acquisition module data acquisition module 2 1 Acquisition motion information analyzing unit 23 and the image feature extraction unit 21, motion data acquisition module 2 for analysis and processing.

[0071] 运动信息解算单元23对所获数据进行解算并将结果传送至多传感器数据融合单元M ;图像特征提取单元21对所获视频图像进行二值算法得到运动目标的特征构图,三维空间定位单元22通过运算变换,得到运动目标的三维空间坐标,并将其传送至多传感器数据融合单元M ;多传感器数据融合单元M将得到的数据信息采用基于DS证据理论的多传感器数据融合算法对运动目标的惯性参数和三维空间坐标作处理,获取目标的动作模式数据。 [0071] the motion information obtained solver unit 23 data solver and transfers up to sensor data fusion unit M; image feature extraction unit 21 acquired video image binary algorithm of moving target feature patterned, three-dimensional space positioning unit 22 by calculation transformation, the three-dimensional coordinates of the moving object, and transfers up to sensor data fusion unit M; data multisensor data fusion unit M will be obtained using the multi-sensor data fusion algorithm based on DS theory of motion inertia parameters and three-dimensional coordinates of the target for processing, target acquisition operation mode data.

[0072] 辨识与训练模块3,通过动作模式数据采集模块31采集各种动作模式样本数据, 由训练模块32对样本模式数据进行预处理,得到训练数据。 [0072] Identification and Training Module 3, the operation mode by collecting a variety of data acquisition module 31 sample data mode operation, preprocessed by the training module 32 sample data mode to obtain training data. 图4中的分类器从训练数据中提取反映数据本质特征的特征向量并根据特征向量对其进行分类,建立从特征向量到所属类别之间映射关系的分类器;辨识单元33对经运动数据获取模块2分析得到的待检测的动作模式数据进行预处理,得到辨识数据,从辨识数据中提取特征向量并输入到图4所述分类器中,分类器根据其特征向量进行判别,得到对待辨识动作模式数据的辨识结果。 Figure 4 classifier training data extracted from the data reflect the essential characteristics of a feature vector and feature vectors according to its classification, the mapping between the feature vector belongs to the category of classification; identification unit 33 via the motion data acquisition action to be detected pattern data analysis module 2 was subjected to pre-treatment to obtain identification data extracted from identification feature vector data and input to the FIG. 4 the classifier, the classifier to discriminate based on their feature vectors, we are treated to identify action pattern recognition result data.

[0073] 自定义不同类别的标准动作模式,建立动作模式标准库41,有:如乒乓球的基本动作、网球基本动作、高尔夫挥杆动作、保龄球掷球动作、跑步的抬腿动作等以体育运动项目的基本动作为主的特定体育运动动作模式标准库45,如鞍马运动员基本动作、体操、体育教学动漫游戏类应用的动作等根据体育教学要求定制特殊应用的非特定动作模式标准库46,以及基于特定动作模型库和非特定模型库为辅助体育教学特定功能而设计的由系列动作组成的续动作模型库47,包括体操、舞蹈、武术、健身气功等一套完整的动作。 [0073] Custom different categories of standard operation mode, the operation mode to establish the standard library 41, are: basic movements such as table tennis, tennis basic movements, golf swing, throw a bowling action, running and other sports leg operation the basic operation of sports-based sport-specific mode of action 45 standard library, such as horse athletes basic movements, gymnastics, physical education class animation and game applications and other custom application-specific action in accordance with the requirements of non-specific action sports teaching mode standard library 46, as well as specific actions based on non-specific model libraries and model library Athletics Teaching with specific functions designed by the continued action model library series consisting of 47 actions, including gymnastics, dance, martial arts, Qigong a complete set of actions.

[0074] 将标准动作模式和体育运动教学步骤自定义为教学方案,并输入虚拟体育教学环境中;将辨识结果输入到虚拟教学环境43中,实现人机交互运动状态;以动作标准模式库为参照标准,由动作解析单元42对辨识结果进行解析,得到动作修正方案(如位置、速度、 角度等方面),并将结果输出在显示设备44。 [0074] The standard operation mode and sport teaching steps to customize the program for teaching and enter the virtual PE teaching environment; the identification result to the virtual learning environment (43), of interacting with the state of motion; in standard operation mode library reference standard, parsed by the parsing operation unit 42 identification result, correcting the operation program (e.g., position, speed, angle, etc.), and outputs the result on the display device 44.

[0075] 由于采用在原来的光学追踪方式下增加了惯性追踪方式其综合后使反映目标运动情况,有效扩大了追踪范围,提高了测量的精度,解决了惯性追踪无法获取目标整体信息、不能做复杂的运动识别、敏感性差的问题,同时也解决光学追踪技术还原真实性差、有效追踪范围小以及阻挡影响的问题。 [0075] As a result in the original optical tracking method increases the inertia track mode after its comprehensive so reflect the objectives movement, effectively expanding the tracking range, improve the measurement accuracy, to solve the inertia track can not get goals overall information and can not do complex motion recognition, poor sensitivity of the issue, but also solve the optical tracking technology to restore the authenticity of the poor, small-scale and effective tracking of barrier impacts. 本发明还具有很强的实用性,为体育教学提供一种全新的教学模式,让体育运动教学方法趋于数字化、多媒体化和科学标准化。 The present invention is also highly practical, to provide a new teaching model of physical education, sports teaching methods tend to make the digital, multimedia and scientific standardization.

Claims (10)

1. 一种基于动作识别技术的体育教学辅助系统,用于PC或嵌入式主机,其包括运动数据采集模块、运动数据获取模块、辨识与训练模块、虚拟教学环境模块组成,所述的运动数据采集模块进一步包括:特定波长点光源、不少于二个摄像头; 所述的运动数据获取模块进一步包括:图像特征提取单元、三维空间定位单元、运动信息解算单元、多传感器数据融合单元;所述的辨识与训练模块进一步包括:动作模式数据采集模块、训练模块、辨识单元; 所述的虚拟教学环境模块包括一动作模式标准库、一动作解析单元、一虚拟教学环境、 一显示设备;其特征在于:在所述的运动数据采集模块中还是设有一微惯性测量单元和一惯性参数提取单元,微惯性测量单元与被测目标绑定,用于测量被测目标的惯性参数,供运动数据获取模块处理,所述的微惯性测量单元通过无线信号与所述的惯性参数提取单元信号输入端连接;所述的运动信息解算单元对惯性参数提取单元输出的数据传输到多传感器数据融合单元作解算处理。 A motion recognition technology based on physical education support system for PC or embedded host, which includes motion data acquisition module, motion data acquisition module, identification and training module, virtual learning environment modules, according to the motion data acquisition module further comprises: a specific wavelength of light source, at least two cameras; the motion data acquisition module further comprises: an image feature extraction unit, three-dimensional positioning unit, the motion information solver unit, multi-sensor data fusion unit; the identification and said training module further comprises: an operation mode data acquisition module, training module identification unit; the virtual learning environment module comprises a standard library operation mode, a motion analysis unit, a virtual learning environment, a display device; it characterized in that: said motion data acquisition module is provided with a micro inertial measurement unit and an inertial parameter extraction unit, micro inertial measurement unit and the measured binding targets for measuring the measured object inertia parameters for motion data Get processing module, the micro inertial measurement unit extraction unit signal input terminal is connected via a wireless signal and the inertial parameters; the motion information of the resolver unit inertial parameter extraction data transmission unit output to the multi-sensor data fusion unit for solving process.
2.如权利要求1所述的一种基于动作识别技术的体育教学辅助系统,其特征在于:所述的运动数据获取模块,其中所述的三维空间定位单元利用标记点在两个摄像头内成像的位置,利用双目视觉算法得到标记点的三维空间坐标,所述多传感器数据融合单元采用基于DS证据理论的多传感器数据融合算法对运动目标的惯性参数和三维空间坐标作处理, 获取目标的动作模式数据。 2. An claim 1, wherein the motion recognition technology based on physical education support system, characterized in that: said motion data acquisition module, wherein the three-dimensional positioning unit utilizes two markers in the camera image position, binocular vision algorithms to obtain three-dimensional coordinates of markers, the multi-sensor data fusion unit based on multi-sensor data fusion DS evidence theory algorithm for three-dimensional movement of inertia parameters and coordinates of the target for treatment, to obtain the target operation mode data.
3.如权利要求1所述的一种基于动作识别技术的体育教学辅助系统,其特征在于:所述的虚拟教学环境模块,其中:所述动作模式标准库用于供动作解析单元解析动作模式的范本;所述的动作模式标准库包括特定体育运动动作模式标准库、非特定动作模式标准库和连续动作模型库; 所述动作解析单元用于解析动作模式数据的辨识结果; 所述虚拟教学环境为人机互动平台,用于提供体育教学方案; 所述显示设备用于反馈运动动作模式辨识的解析结果、显示人机交互动作状态,其输入端与计算机连接;所述的特定体育运动动作模式标准库,以体育运动项目的基本动作为主,包括乒乓球的基本动作、网球基本动作、高尔夫挥杆动作、保龄球掷球动作、跑步的抬腿动作等;所述的非特定动作模式标准库,是根据体育教学要求定制特殊应用的动作模型库,包括鞍马运动员基本动作、体操、体育教学动漫游戏类应用的动作等;所述的连续动作模型库,为辅助体育教学特定功能而设计的由系列动作组成的组合动作库,它基于特定动作模型库和非特定模型库,包括体操、舞蹈、武术、健身气功等一套完整的动作。 3. The one claim 1, wherein the motion recognition technology based on physical education assistance system, wherein: the virtual learning environment module, wherein: the operation mode of the standard library for parsing unit for parsing operation mode of action template; the operation mode of the standard library includes a specific sport operation mode of the standard library, non-specific mode of action of the standard library and continuous action model library; motion analysis unit for identifying the result of analysis of the operation mode data; the virtual Teaching environment for human-machine interactive platform for providing sports education programs; the analysis result display apparatus for motion feedback operation mode identification, display human-computer interaction operation state, having an input connected to the computer; the specific mode of action sports standard library to basic operation of sports-based projects, including the basic movements of table tennis, tennis basic movements, golf swing, throw a bowling action, running leg movement, etc; non-specific mode of action of the standard library , based on PE teaching customized application-specific action model library, including the pommel horse athletes basic movements, gymnastics, physical education class animation and game applications and other actions; continuous action of the model base for the Athletics Teaching with specific features designed by series of actions consisting of a combination of action library, which is based on specific and non-specific action model library model database, including gymnastics, dance, martial arts, Qigong a complete set of actions.
4. 一种基于动作识别技术的虚拟体育教学辅助系统的构建方法,其包括以下步骤:A、将标记点与被测目标绑定,将摄像头安装在指定区域的正前方,用于捕捉指定区域的图像、获取标记点运动的惯性参数;B、采集数字图像并输入通用计算机,得到目标运动的数字图像;采集标记点运动的惯性参数;C、对采集的惯性参数和数字图像进行解算和多传感器融合,获取标记点运动方式;D、根据标记点运动方式识别其运动模式,传送给虚拟体育教学环境,并反馈输出相应的交互动作状态,对输出的交互动作状态进行统计,并与预设的体育教学指标进行比对和解析,得到动作修正方案如位置、速度、角度等方面,并将结果输出在显示设备上。 4. A construction method based on motion recognition technology virtual PE teaching assistant system, comprising the steps of: A, the markers and the measured binding target, install the camera in front of the designated area for capturing the designated area image Get inertia parameter markers movement; B, collecting digital images and enter a general purpose computer, get a digital image of the target movement; inertia parameter acquisition marker movement; C, inertia parameters and digital image acquisition were solver and multi-sensor fusion, get marker movement; D, according to the marker movement identify its Sport mode, the transmission to the virtual PE teaching environment, and the feedback output corresponding interactive operation state, the output of the interaction state statistics, and pre Physical education indicators is provided for comparison and analysis, to obtain correction action plan such as position, speed, angle, etc., and outputs the result on a display device.
5.如权利要求4所述的基于动作识别技术的虚拟体育教学辅助系统的构建方法,其特征在于:所述的步骤C还包括:在图像特征提取单元,首先通过二值算法提取目标运动的特征构图,对特征构图求取其重心,得到特征构图中心位置;接着根据对所选特征的计算,建立特征之间的对应关系; 5. The construction of auxiliary systems based on virtual technology of action recognition PE teaching method according to claim 4, wherein: said step C further comprising: an image feature extraction means for extracting first target motion by binary arithmetic characteristic composition, its center of gravity to strike the characteristic composition to give the composition features a central location; then calculated for the selected feature, to establish correspondence between the features;
6.如权利要求4所述的基于动作识别技术的虚拟体育教学辅助系统的构建方法,其特征在于:所述步骤C还包括:对所述步骤B中获取的标记点运动方式和惯性参数,通过DS 证据理论的多传感器数据融合算法,得到反映目标运动特征的一致性数据。 6. The method of motion recognition technology based build virtual PE Teaching System according to claim 4, wherein: said step C further comprising: a marker motion and inertia parameters of the step B acquired, multi-sensor data fusion algorithm DS evidence theory, target motion characteristics reflected the consistency of the data.
7.如权利要求4所述的基于动作识别技术的虚拟体育教学辅助系统的构建方法,其特征在于:所述步骤D还包括:D1、采集各种样本动作模式数据,对采集到的样本动作模式数据进行标注; D2、逐一从所述样本动作模式数据中提取出反映其本质特征的特征向量; D3、根据所述特征向量划分所属类别区域,使得划分后的各个不同类别区域中只包含同类样本的特征向量,建立从特征向量到所属类别之间映射关系的分类器; D4、对待辨识动作模式数据进行处理,提取其特征向量;D5、将待辨识动作模式数据的特征向量输入到所述分类器,分类器根据其特征向量进行判别,得到对该待辨识动作模式数据的辨识结果。 7. The method of motion recognition technology based build virtual PE Teaching System according to claim 4, wherein: said step D further comprises: D1, various sample collection operation mode data, the collected samples action mode data tagging; D2, extracted one by one to reflect its essential characteristics of a feature vector from the sample data in operation mode; the various categories of regions D3, according to the feature vector divided category region so that the divided contains only the same sample feature vectors, the mapping between feature vectors to the category for which the classifier; the D4, treat operation mode identification data is processed to extract eigenvectors; D5, the motion pattern data to be recognized is input to the feature vectors classifier, classifier judgment is made based feature vectors give the motion pattern data to be recognized in the recognition result. D6、动作模式数据的辨识结果输入到虚拟教学环境和动作解析单元中,通过显示设备, 反馈输出运动状态。 The D6, the operation mode of the recognition result data input to the virtual environment and the operation instruction parsing unit, through the display device, the output of the feedback motion.
8.如权利要求7所述的基于动作识别技术的虚拟体育教学辅助系统的构建方法,其特征在于:所述步骤D2还包括:D21、对所述样本动作模式数据进行预处理,得到训练数据; D22、从训练数据中提取反映训练数据本质特征的特征成分; D23、将所述特征成分进行组合,得到所述特征向量。 8. The method of motion recognition technology based build virtual PE Teaching System according to claim 7, wherein: the step D2 further comprises: D21, the operation mode of the sample data preprocessing to obtain training data ; D22, extracted from the training data reflect the essential characteristics of the training data characteristic components; D23, the characteristic ingredients are combined to obtain the feature vector.
9.如权利要求7所述的基于动作识别技术的虚拟体育教学辅助系统的构建方法,其特征在于:所述步骤D4还包括:D41、对所述待辨识动作模式数据进行预处理,得到辨识数据; D42、从辨识数据中提取反映辨识数据本质特征的特征成分; D43、将所述特征成分进行组合,得到所述特征向量。 9. The method of action based on recognition technology to build virtual PE Teaching System according to claim 7, wherein: said step D4 further comprises: D41, suspicious of the operation mode data preprocessing to obtain recognition data; D42, extracted from the data identify the identification data reflect the essential characteristics of the characteristic components; D43, the characteristic ingredients are combined to obtain the feature vector.
10.如权利要求7所述的基于动作识别技术的虚拟体育教学辅助系统的构建方法,其特征在于:所述步骤D6还包括:D61、根据辨识数据的特征向量,自定义不同类别的标准动作模式,建立动作模式标准库;D62、将标准动作模式和体育运动教学步骤自定义为教学方案,并输入虚拟体育教学环境中;D63、将辨识结果输入到虚拟体育教学环境中,现实人机交互运动状态;以标准动作库为参照标准,对辨识结果进行解析,得到动作修正方案;D64、将动作修正方案和动作模式数据辨识结果输入到显示设备。 10. The claim 7 based virtual sports teaching aids movement recognition technology to build a systematic approach, characterized by: further comprising the step D6: D61, based on identification feature vector data, customized standard action different categories mode, the establishment of an operation mode of the standard library; D62, the standard operation mode and sport teaching steps to customize the program for teaching and enter the virtual PE teaching environment; D63, the recognition result is input to the virtual physical education teaching environments, realistic human-computer interaction motion; a standard action library as a reference standard analytical results of identification to obtain a correction action plan; D64, the correction action plan and action pattern data recognition result input to the display device.

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2013104316A1 (en) * 2012-01-09 2013-07-18 Zienon Electronic Technology (Xi'an) Co., Ltd. Method and device for filter-processing imaging information of emission light source
CN103212192A (en) * 2012-01-24 2013-07-24 精工爱普生株式会社 Motion analysis system and motion analysis method
CN104066483A (en) * 2012-01-31 2014-09-24 高夫准株式会社 Virtual golf simulation apparatus and method for supporting generation of virtual green
CN102989126A (en) * 2012-11-23 2013-03-27 安徽寰智信息科技股份有限公司 Action recognition-based child hand and eye coordination training and evaluating system
CN102989163A (en) * 2012-11-23 2013-03-27 安徽寰智信息科技股份有限公司 Action recognition-based bowling bowl training auxiliary system
CN102989121B (en) * 2012-11-23 2015-06-24 安徽寰智信息科技股份有限公司 Action recognition-based Alzheimer's disease preventing and rehabilitation training system
CN102989126B (en) * 2012-11-23 2016-03-09 安徽寰智信息科技股份有限公司 Children's hand-eye coordination training and evaluation system based on motion recognition
CN102989121A (en) * 2012-11-23 2013-03-27 安徽寰智信息科技股份有限公司 Action recognition-based Alzheimer's disease preventing and rehabilitation training system
CN103106411A (en) * 2012-12-13 2013-05-15 徐玉文 Tennis motion capturing and analyzing method
CN103176606B (en) * 2013-04-15 2016-03-16 上海图漾信息科技有限公司 Flat-screen interactive system and method based on binocular visual identity
CN103176606A (en) * 2013-04-15 2013-06-26 北京唯创视界科技有限公司 Plane interaction system and method based on binocular vision recognition
CN103456197A (en) * 2013-09-18 2013-12-18 重庆创思特科技有限公司 Wireless intelligent somatosensory interaction front end for teaching
CN103578135A (en) * 2013-11-25 2014-02-12 恒德数字舞美科技有限公司 Virtual image and real scene combined stage interaction integrating system and realizing method thereof
CN103578135B (en) * 2013-11-25 2017-01-04 恒德数字舞美科技有限公司 Stage interactive systems integration and realization of virtual images and real scene combining

Similar Documents

Publication Publication Date Title
Brostow et al. Semantic object classes in video: A high-definition ground truth database
Xia et al. View invariant human action recognition using histograms of 3d joints
Andriluka et al. People-tracking-by-detection and people-detection-by-tracking
Yang et al. Recognizing actions using depth motion maps-based histograms of oriented gradients
Han et al. Enhanced computer vision with microsoft kinect sensor: A review
Luber et al. People tracking in rgb-d data with on-line boosted target models
US20120070070A1 (en) Learning-based pose estimation from depth maps
Gavrila et al. Vision-based pedestrian detection: The protector system
Zanfir et al. The moving pose: An efficient 3d kinematics descriptor for low-latency action recognition and detection
US8166421B2 (en) Three-dimensional user interface
US7036094B1 (en) Behavior recognition system
Bobick et al. The recognition of human movement using temporal templates
US20120062736A1 (en) Hand and indicating-point positioning method and hand gesture determining method used in human-computer interaction system
Chambers et al. Hierarchical recognition of intentional human gestures for sports video annotation
Ji et al. Advances in view-invariant human motion analysis: A review
US20090220124A1 (en) Automated scoring system for athletics
Sun et al. Hierarchical spatio-temporal context modeling for action recognition
US20090060352A1 (en) Method and system for the detection and the classification of events during motion actions
Yun et al. Two-person interaction detection using body-pose features and multiple instance learning
Chaquet et al. A survey of video datasets for human action and activity recognition
Lai et al. A gesture-driven computer interface using Kinect
Ryoo et al. An overview of contest on semantic description of human activities (SDHA) 2010
US20080285807A1 (en) Apparatus for Recognizing Three-Dimensional Motion Using Linear Discriminant Analysis
Bloom et al. G3d: A gaming action dataset and real time action recognition evaluation framework
CN102426480A (en) Man-machine interactive system and real-time gesture tracking processing method for same

Legal Events

Date Code Title Description
C06 Publication
C02 Deemed withdrawal of patent application after publication (patent law 2001)