CN102096812A - Teacher blackboard writing action detection method for intelligent teaching recording and playing system - Google Patents

Teacher blackboard writing action detection method for intelligent teaching recording and playing system Download PDF

Info

Publication number
CN102096812A
CN102096812A CN2011100323381A CN201110032338A CN102096812A CN 102096812 A CN102096812 A CN 102096812A CN 2011100323381 A CN2011100323381 A CN 2011100323381A CN 201110032338 A CN201110032338 A CN 201110032338A CN 102096812 A CN102096812 A CN 102096812A
Authority
CN
China
Prior art keywords
detection
image
color
motion
background
Prior art date
Application number
CN2011100323381A
Other languages
Chinese (zh)
Inventor
吴柯维
王凤宇
许松涛
贾子杰
魏周朝
Original Assignee
吴柯维
王凤宇
许松涛
贾子杰
魏周朝
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
Application filed by 吴柯维, 王凤宇, 许松涛, 贾子杰, 魏周朝 filed Critical 吴柯维
Priority to CN2011100323381A priority Critical patent/CN102096812A/en
Publication of CN102096812A publication Critical patent/CN102096812A/en

Links

Abstract

The invention discloses a teacher blackboard writing action detection method for an intelligent teaching recording and playing system. The method comprises the following steps of: acquiring and marking an image; acquiring a frame of image by using a video acquisition card; calibrating a blackboard region which is in a closed polygon; detecting motion through background modeling and updating, background differencing, skin color differencing, motion detection and motion combination; detecting a region by methods such as an area method, a pixel statistical method, wherein the pixel statistical method comprises the following step of: calculating a valid binary number in a detection region, a calculation formula is shown in the specifications, and the area method comprises the following steps of: searching an Ibin communicated region; and calculating the total area s of the region; detecting blackboard writing; and if s is more than a certain threshold value, recognizing the blackboard writing action of a teacher. The method has high correct blackboard writing action detection rate.

Description

一种教学智能录播系统用教师黑板书写动作检测方法 A teaching intelligent motion detection recording and broadcasting system written in the blackboard teacher

技术领域 FIELD

[0001] 本发明涉及一种教学智能录播系统用教师黑板书写动作检测方法,适用于现代信息化教室、智能课堂教学录播系统、精品课程录制、交互式教学系统的建设。 [0001] The present invention relates to a method of teaching intelligent motion detection recording and broadcasting system written in teacher blackboard for a classroom of modern information technology, construction of classroom teaching intelligent recording and broadcasting system, recording quality courses, interactive teaching systems.

背景技术 Background technique

[0002] 随着教育信息化的发展,多媒体课件在教学过程中应用的日益广泛,越来越多的学校建成或在建电子信息化教室(简称电教室),电教室不但能进行多媒体教学,还有专人将教学过程同步录制下来,形成学校课堂教学录像。 [0002] With the development of information technology in education, multimedia courseware in the teaching process applications increasingly widespread, more and more schools built or under construction, electronics and information technology classroom (referred to as the classroom), not only classroom multimedia teaching, there is someone teaching process will synchronize recorded, forming a school classroom teaching videos. 目前的课堂教学录像,不但有课件拍摄,还有学生起立回答问题、教师黑板板书、教师讲解问题的拍摄,因为拍摄角度的不同,势必需要多个机位摄像机参与拍摄,多机位的拍摄需要有导播来完成各摄像机之间的镜头转场、特技等现场编辑,这就需要众多的摄像人员和一位导播员。 The current classroom teaching video, not only courseware shooting, as well as the students stood up to answer questions, the teacher blackboard writing on the blackboard, teachers on the issue of the shooting, because of the different camera angles, will be needed more seats participated in the filming camera, multi-camera shooting needs there are directed to complete the lens between the camera transitions, special effects and other editing site, which requires a large number of camera crews and directed staff. 随着国家“精品课程”建设项目推广,以及传统的教学评估、示范课录制、教学观摩、远程教学等建设,无一不需要课堂教学录像,为了将管理电教室的教师从繁重的录播任务解放出来,自动录播系统出现了,能够自动进行多机位拍摄,自动导播,录像录制和教师讲课同步进行,课结束了,多媒体课堂教学录像自动完成了。 With the national "Quality Courses" to promote the construction project, as well as traditional teaching assessment, recording the construction of demonstration lessons, classroom observation, distance learning, and both require classroom teaching video, in order to manage classroom teacher from the arduous task of taped liberation, automatic recording and broadcasting system emerged, capable of automatic multi-camera shoot, automatically directed, teacher lectures and video recording synchronization, the end of the lesson, multimedia classroom teaching video done automatically.

[0003] 目前教师黑板书写动作检测,有两种方法,一种是采用被动式红外探测技术,将一个红外发射装置安装在教师的腰部,当教师转身写板书时,红外接收器接受到信号,以别为教师写板书;另一种方法是通过视频帧间差分分析,如果人的胳膊位于检测出目标的左上或右上,则认为是有写板书动作目前的教师位置检测方法,有基于光学红外发光阵列定位、 主动式超声波探测技术、教师佩戴发射器定位技术、基于运动检测的多通道融合技术等方法。 [0003] It blackboard writing teachers motion detection, there are two methods, one is the use of passive infrared detection technology, the infrared emitting device is mounted on a teacher's waist, written on the blackboard when the teacher turned, the receiver receives the infrared signal to do not write the blackboard for teachers; another method is by differential analysis of video frames, if the human arm is located in the upper left or upper right target is detected, it is considered to be the current position detection method teacher blackboard writing operation, there is an optical based on infrared light emitting array is positioned, the active ultrasonic detection technology, teacher wearing a transmitter location technology, motion detection based on multi-channel fusion technology. 其中基于光学红外发光阵列定位方法是在黑板的下方,安装一系列的光学发光阵列,在教师后方,安装有教师定位仪,通过判断教师遮挡发光阵列的位置来定位教师所在的位置。 Wherein the infrared light emitting optical array is positioned below based on the blackboard, a series of mounting an optical light emitting array, in the rear teacher mounted teacher locator to locate the position where the teacher is determined by the position of the light emitting array teacher blocked.

[0004] 被动式红外探测技术缺点是抗干扰能力较差,只要教师一转身,就被识别为写板书动作,这无疑导致自动导播的误切换。 [0004] Passive infrared detection technology disadvantage is poor anti-interference ability, as long as the teacher turned around, he was identified as written on the blackboard action, which will undoubtedly lead to false switching automatically directed. 该视频帧间差分分析技术的缺点是,如果教师背对黑板,手臂向上挥动时,会被识别为写板书动作,这也是该方法固有的缺陷,不可避免的造成误识别,影响最终录播结果,而且,当教师手放在黑板上,不动时,不会识别。 The disadvantage of this technical analysis video frame difference is that if a teacher back blackboard, when the arm swing upwards, will be recognized as written on the blackboard action, which is the method inherent flaws inevitably cause false identification, recording and broadcasting affect the final result Moreover, when teachers hand on the blackboard, do not move, it does not recognize.

发明内容 SUMMARY

[0005] 为解决上述问题,本发明提供一种正确一种教学智能录播系统用教师黑板书写动作检测方法 [0005] In order to solve the above problems, the present invention provides a broadcasting system correctly written in an intelligent motion detection teacher teaching blackboard

[0006] 本发明实现上述目的的技术方案是:一种教学智能录播系统用教师黑板书写动作检测方法,其创新点在于:其步骤包括, [0006] The object of the present invention to achieve the above technical solution is: A method for detecting an operation instruction written in the intelligent broadcasting system teacher blackboard innovative in that: comprising the steps of,

[0007] 1)图像采集和标定:使用视频采集卡,采集一帧图像;对黑板区域进行标定,闭合多边形内是黑板区域; [0007] 1) image acquisition and calibration: using a video capture card, one frame of image capture; calibration area on the board, the board area is a closed polygon;

[0008] 2)运动检测:包括通过背景建模与更新、背景差分、肤色差分、运动检测、运动合并进行运动检测;[0009] 3)区域检测:采用有面积法和像素统计法等方法,像素统计法为:计算落在检测区域内有效二值个数,计算公式如下 [0008] 2) Motion Detection: by comprising update background modeling and background difference, color difference, motion detection, motion detection combined motion; [0009] 3) Detection region: the method has an area of ​​France and the pixel statistics, method for the pixel statistics: the number of valid binary calculation falls within the detection region is calculated as follows

[0010] [0010]

Figure CN102096812AD00051

[0011] 面积法为:搜索Ibin联通的区域,然后计算区域总面积s ; [0011] Area law: Ibin Unicom search area, and then calculating the total area of ​​the region S;

[0012] 4)板书检测:若s大于一定的阈值后,则识别为教师写板书动作。 [0012] 4) detection blackboard: If after s is greater than a certain threshold, it is recognized that operation of the blackboard writing teachers.

[0013] 所述图像采集和标定是通过在黑板边框四周及讲台顶部安装的摄像机采集图像。 [0013] The image acquisition and calibration images are acquired by a camera on top of the board and the platform is installed around the border.

[0014] 所述背景建模与背景更新的方法为:令/Λ"(Χ,力代表背景帧图像,初始化为第一帧,η表示当前时刻,(x,y)代表像素的位置,厂(W)表示当前帧图像,渐进均值滤波背景建模可以用如下公式表示: [0014] The background modeling and updating method: Order / Λ "(Χ, a force representative of the background frame image is initialized to a first frame, [eta] represents the current time, (x, y) position of the representative pixels, plant (W) indicates the current frame image, background modeling progressive mean filter can be formulated as follows:

[0015] [0015]

Figure CN102096812AD00052

[0016] 均值滤波背景建模可以用如下公式表示: [0016] BACKGROUND mean filter model can be expressed by the following formula:

[0017] [0017]

Figure CN102096812AD00053

[0018] 中值滤波背景建模可以用如下公式表示: [0018] background modeling median filtering can be expressed by the following formula:

[0019] [0019]

Figure CN102096812AD00054

[0020] 近似中值滤波背景建模可以用如下公式表示: [0020] The approximate background modeling median filtering can be expressed by the following formula:

[0021] [0021]

Figure CN102096812AD00055

[0022] 背景差分的方法为:令Ib(X,y)代表背景帧图像,Ic(x, y)表示当前帧图像,Id(x, y)表示差分图像,差分公式如下: [0022] The background difference method: Let Ib (X, y) representative of the background frame image, Ic (x, y) represents the current frame image, Id (x, y) represents a differential image, the differential equation is as follows:

[0023] Id(X,y) = Ic(x, y)_Ib(x,y) | ; [0023] Id (X, y) = Ic (x, y) _Ib (x, y) |;

[0024] 运动检测的过程为:表示将差分图像根据一阈值,划分为运动与非运动像素,运动二值图像Ibin(x,y)可以只有0与1,也可以只有0与255,也可以是任意不同的两个值组成, Process [0024] The motion detection is: shows a differential image according to a threshold value, is divided into the motion and non-motion pixels, the motion binary image Ibin (x, y) can be 0 and 1, may be only 0 and 255, may be are any two different values ​​of the composition,

公式如下: Formula is as follows:

[0025] [0025]

Figure CN102096812AD00056

[0026] 肤色检测,目前常见肤色检测检测中采用了各种各样的颜色空间,有RGB、HSV、 YCbCr与Lab等,肤色检测方法有基于非参数模型、基于参数模型、基于支持向量机(SVM) 模型、基于神经网络模型、基于红外感知的肤色检测、基于近红外图像和基于高斯分布等方法。 [0026] skin-color detection, skin color detection current common detection using a variety of color space, RGB, HSV, YCbCr Lab and the like, with a non-parametric method of detecting skin color model, based on the parameter model, support vector machine ( SVM) model, neural network model based on infrared detection based on skin color perception, based on near-infrared images and methods based on Gaussian distribution. 以上方法适用于本发明但不局限于以上方法,只要用到了人体肤色检测,均适用于本步骤。 The method of the present invention is more suitable for, but not limited to the above method, as long as the human skin used in the detection, are applicable to this step. 令Is(x,y)为肤色图像,I(x,y)为红外图像,则肤色检测公式如下: Order Is (x, y) is a color image, I (x, y) is an infrared image, the color detection formula is as follows:

[0027] [0027]

Figure CN102096812AD00057

[0028] 基于高斯分布的肤色检测方法步骤是,在YCbCr空间,对不同图像进行肤色采样, 计算样本的均值M和协方差C,令P表示概率值,则肤色检测为 [0028] Skin Color Detection step based on the Gaussian distribution, in the YCbCr space, sampling of the different color images, calculate the sample mean and covariance C M, P denotes the probability value so that, as the skin color detection

[0029] [0029]

Figure CN102096812AD00061

[0030] 肤色差分,表示使用当前帧肤色图减去背景帧肤色图,但不局限于使用当前帧,也可以使用当前时刻前后相邻若干时刻帧的图像,只要使用了背景帧肤色做减法,均在肤色差分范畴。 [0030] The color difference, using the current frame by subtracting a background frame in FIG skin color chart, but not limited to the use of the current frame, a plurality of images may be used before and after the current time frame next time, as long as the use of the color of the background frame subtraction, They are in the category of color difference. 令Ibs(χ,y)代表背景帧肤色图像,Ics(χ,y)表示当前帧肤色图像,Ids(x, y)表示肤色差分图像,差分公式如下: Order Ibs (χ, y) representative of the background color image frames, Ics (χ, y) represents the color image of the current frame, Ids (x, y) represents color difference image, difference formula as follows:

[0031 ] Ids (x, y) = I Ics (X,y) -Ibs (χ, y) [0031] Ids (x, y) = I Ics (X, y) -Ibs (χ, y)

[0032] 肤色运动检测与运动检测步骤相同; [0032] The color motion detection and motion detection step of the same;

[0033] 运动检测合并:即4„0^)二4„(^)«4>,力,@表示二进制与,I' bin为帧间差 [0033] The combined motion detection: i.e. 4 "^ 0) = 4" (^) «4>, force, and represents a binary @, I 'bin for the inter-frame difference

分检测结果,Ibin为背景差分运动检测结果,并存放合并后检测结果。 Min detection result, Ibin background differential motion detection result and the detection result of the combined store.

[0034] 本发明的有效效果为:本发明将摄像机检测黑板技术、背景建模技术、背景差分技术、运动检测技术、肤色检测技术运用到了教室写板书动作识别中,与被动式红外探测技术相比,发明不用在教师身上佩戴任何设备,在不影响教师正常讲课的情况下,消除了教师转身造成的误判;与视频帧间差分分析技术相比,本发明在教师背对黑板,手臂向上挥动时, 不会产生误报。 [0034] The active effect of the present invention are: the present invention will be detected by the camera blackboard art, background modeling technique, background subtraction techniques, motion detection, skin-color detection technique applied to the classroom write blackboard action recognition as compared to the passive infrared detection technology , teachers who invention without wearing any equipment, without affecting the normal teacher lectures, eliminating the false teachers turn caused; compared with the video frame difference analysis techniques, the present invention teacher blackboard with his back, arms waving up when, without producing false alarms. 本发明在正确写板书动作检测率方面远远好于前两款技术。 In the present invention, the right to write on the blackboard motion detection rates far better than the previous two techniques.

附图说明 BRIEF DESCRIPTION

[0035] 图1是本发明的流程图; [0035] FIG. 1 is a flow chart of the present invention;

[0036] 图2是本发明的摄像机位置示意图。 [0036] FIG. 2 is a schematic view of the present invention the camera position.

具体实施方式 Detailed ways

[0037] 下面结合说明书附图对本发明作进一步说明。 [0037] the following description in conjunction with the accompanying drawings of the present invention will be further described.

[0038] 如图1所示,一种教学智能录播系统用教师黑板书写动作检测方法,其步骤包括, [0038] As shown in FIG. 1, a teaching method for intelligent motion detection teacher broadcasting system written in the blackboard, comprising the steps of,

[0039] 1)图像采集和标定:使用视频采集卡,采集一帧图像;对黑板区域进行标定,闭合多边形内是黑板区域; [0039] 1) image acquisition and calibration: using a video capture card, one frame of image capture; calibration area on the board, the board area is a closed polygon;

[0040] 2)运动检测:包括通过背景建模与更新、背景差分、肤色差分、运动检测、运动合并进行运动检测; [0040] 2) Motion Detection: by comprising update background modeling and background difference, color difference, motion detection, motion detection combined motion;

[0041] 3)区域检测:采用有面积法和像索统计法等方法,像素统计法为:计算落在检测区域内有效二值个数,计算公式如下 [0041] 3) Detection region: The area method with a cable and an image statistics methods such as the pixel statistics: the number of calculation fall within a valid binary detection area, calculated as

[0042] [0042]

Figure CN102096812AD00062

[0043] 面积法为:搜索Ibin联通的区域,然后计算区域总面积s ; [0043] Area law: Ibin Unicom search area, and then calculating the total area of ​​the region S;

[0044] 4)板书检测:若s大于一定的阈值后,则识别为教师写板书动作。 [0044] 4) detection blackboard: If after s is greater than a certain threshold, it is recognized that operation of the blackboard writing teachers.

[0045] 如图2所示,所述图像采集和标定是通过在黑板边框四周及讲台顶部安装的摄像机采集图像。 [0045] As illustrated, the image acquisition and calibration images are acquired through the camera 2 around the border at the top of the board and the mounting platform. [0046] 所述背景建模与背景更新的方法为:令《(U)代表背景帧图像,初始化为第一帧,η表示当前时刻,(x,y)代表像素的位置,^(U)表示当前帧图像,渐进均值滤波背景建模可以用如下公式表示: [0046] The background modeling and updating method: Order "(U) representative of the background frame image is initialized to a first frame, [eta] represents the current time, (x, y) position of the representative pixels, ^ (U) represents the current frame image, background modeling progressive mean filter can be formulated as follows:

Figure CN102096812AD00071

[0048] 均值滤波背景建模可以用如下公式表示: [0048] BACKGROUND mean filter model can be expressed by the following formula:

Figure CN102096812AD00072

[0050] 中值滤波背景建模可以用如下公式表示: [0051 ] I" (x, y) 二MedianiJ' (χ, >-)) [0050] background modeling median filtering can be expressed by the following equation: [0051] I "(x, y) two MedianiJ '(χ,> -))

Figure CN102096812AD00073

[0052] 近似中值滤波背景建模可以用如下公式表示: [0052] The approximate background modeling median filtering can be expressed by the following formula:

Figure CN102096812AD00074

[0054] 背景差分的方法为:令Ib(X,y)代表背景帧图像,Ic(x, y)表示当前帧图像,Id(x, y)表示差分图像,差分公式如下: [0054] The background difference method: Let Ib (X, y) representative of the background frame image, Ic (x, y) represents the current frame image, Id (x, y) represents a differential image, the differential equation is as follows:

Figure CN102096812AD00075

[0056] 运动检测的过程为:表示将差分图像根据一阈值,划分为运动与非运动像素,运动二值图像Ibin(x,y)可以只有0与1,也可以只有0与255,也可以是任意不同的两个位组成, Process [0056] The motion detection is: shows a differential image according to a threshold value, is divided into the motion and non-motion pixels, the motion binary image Ibin (x, y) can be 0 and 1, may be only 0 and 255, may be is any different two bits,

公式如下: Formula is as follows:

[0057] [0057]

Figure CN102096812AD00076

[0058] 肤色检测,目前常见肤色检测检测中采用了各种各样的颜色空间,有RGB、HSV, YCbCr与Lab等,肤色检测方法有基于非参数模型、基于参数模型、基于支持向量机(SVM) 模型、基于神经网络模型、基于红外感知的肤色检测、基于近红外图像和基于高斯分布等方法。 [0058] skin-color detection, skin color detection current common detection using a variety of color space, RGB, HSV, YCbCr Lab and the like, with a non-parametric method of detecting skin color model, based on the parameter model, support vector machine ( SVM) model, neural network model based on infrared detection based on skin color perception, based on near-infrared images and methods based on Gaussian distribution. 以上方法适用于本发明但不局限于以上方法,只要用到了人体肤色检测,均适用于本步骤。 The method of the present invention is more suitable for, but not limited to the above method, as long as the human skin used in the detection, are applicable to this step. 令Is(x,y)为肤色图像,I(x,y)为红外图像,则肤色检测公式如下: Order Is (x, y) is a color image, I (x, y) is an infrared image, the color detection formula is as follows:

[0059] [0059]

Figure CN102096812AD00077

[0060] 基于高斯分布的肤色检测方法步骤是,在YCbCr空间,对不同图像进行肤色采样, 计算样本的均值M和协方差C,令P表示概率值,则肤色检测为 [0060] Skin Color Detection step based on the Gaussian distribution, in the YCbCr space, sampling of the different color images, calculate the sample mean and covariance C M, P denotes the probability value so that, as the skin color detection

「nncil C'\xM) "Nncil C '\ xM)

[0061] Is(x,y) = P(Cb,Cr) = e 2 [0061] Is (x, y) = P (Cb, Cr) = e 2

[0062] 肤色差分,表示使用当前帧肤色图减去背景帧肤色图,但不局限于使用当前帧,也可以使用当前时刻前后相邻若干时刻帧的图像,只要使用了背景帧肤色做减法,均在肤色差分范畴。 [0062] The color difference, using the current frame by subtracting a background frame in FIG skin color chart, but not limited to the use of the current frame, a plurality of images may be used before and after the current time frame next time, as long as the use of the color of the background frame subtraction, They are in the category of color difference. 令Ibs(χ,y)代表背景帧肤色图像,Ics(χ,y)表示当前帧肤色图像,Ids(x, y)表示肤色差分图像,差分公式如下: Order Ibs (χ, y) representative of the background color image frames, Ics (χ, y) represents the color image of the current frame, Ids (x, y) represents color difference image, difference formula as follows:

7[0063] Ids(x,y) 二|lcs(X,y)_Ibs(X,y) 7 [0063] Ids (x, y) two | lcs (X, y) _Ibs (X, y)

[0064] 肤色运动检测与运动检测步骤相同; [0064] The color motion detection and motion detection step of the same;

[0065] 运动检测合并:即O,y) = IhmO,y)® hmO,y),φ表示二进制与,ι ' bin为帧间差 [0065] The combined motion detection: i.e. O, y) = IhmO, y) ® hmO, y), φ represents the binary, ι 'bin for the inter-frame difference

分检测结果,Ibin为背景差分运动检测结果,并存放合并后检测结果。 Min detection result, Ibin background differential motion detection result and the detection result of the combined store.

Claims (4)

1. 一种教学智能录播系统用教师黑板书写动作检测方法,其特征在于:其步骤包括,1)图像采集和标定:使用视频采集卡,采集一帧图像;对黑板区域进行标定,闭合多边形内是黑板区域;2)运动检测:包括通过背景建模与更新、背景差分、肤色差分、运动检测、运动合并进行运动检测;3)区域检测:采用有面积法和像素统计法等方法,像素统计法为:计算落在检测区域内有效二值个数,计算公式如下Σαά 少)检测区,Ii是运动像素面积法为:搜索Ibin联通的区域,然后计算区域总面积s ;4)板书检测:若s大于一定的阈值后,则识别为教师写板书动作。 An intelligent recording and broadcasting system teaching writing operation detecting method teacher blackboard, characterized in that: comprising the steps of, 1) image acquisition and calibration: using a video capture card, one frame of image capture; calibration area on a blackboard, a closed polygon the blackboard region; 2) motion detection: including background modeling and motion is detected by the update, background difference, color difference, motion detection, motion combined; 3) detection region: there are methods using the method and the pixel area statistics, the pixel statistics to: calculate the number falls within the valid region detected binary, calculated as less Σαά) detection zone, the pixel area of ​​Ii is a motion law: Ibin Unicom search area, and then calculating the total area of ​​the region s; 4) detecting blackboard : If after s is greater than a certain threshold value, the recognition operation of the blackboard writing teachers.
2.根据权利要求1所述的一种教学智能录播系统用教师黑板书写动作检测方法,其特征在于:所述图像采集和标定是通过在黑板边框四周及讲台顶部安装的摄像机采集图像。 The intelligent recording and broadcasting a teaching system according to claim 1 motion detection method using writing teachers blackboard, wherein: the calibration and image acquisition by camera to capture the top of the board and the platform is installed around the border.
3.根据权利要求1所述的一种教学智能录播系统用教师黑板书写动作检测方法,其特征在于:所述背景建模与背景更新的方法为:令代表背景帧图像,初始化为第一帧, η表示当前时刻,U,y)代表像素的位置,^(U)表示当前帧图像,渐进均值滤波背景建模可以用如下公式表示: 均值滤波背景建模可以用如下公式表示: The intelligent recording and broadcasting a teaching system according to claim 1 motion detection method using writing teachers blackboard, wherein: said background modeling and method of updating is: Let frame image representative of the background, is initialized to a first frame, [eta] represents the current time, U, y position) representative pixels, ^ (U) represents the current frame image, background modeling progressive mean filter can be formulated as follows: mean filtering background modeling may be represented by the following formula:
Figure CN102096812AC00021
中值滤波背景建模可以用如下公式表示: Median filtering the background model can be expressed by the following equation:
Figure CN102096812AC00022
近似中值滤波背景建模可以用如下公式表示: Approximate median filtering background modeling may be represented by the following formula:
Figure CN102096812AC00023
背景差分的方法为:令Ib(x,y)代表背景帧图像,I。 The background difference method: Let Ib (x, y) representative of the background frame image, I. (X,y)表示当前帧图像,Id(x,y)表示差分图像,差分公式如下: (X, y) represents the current frame image, Id (x, y) represents a differential image, the differential equation is as follows:
Figure CN102096812AC00024
4.运动检测的过程为:表示将差分图像根据一阈值,划分为运动与非运动像素,运动二值图像Ibin(X,y)可以只有O与1,也可以只有0与255,也可以是任意不同的两个值组成,公式如下: Process 4. The motion detection is: shows a differential image according to a threshold value, is divided into the motion and non-motion pixels, the motion binary image Ibin (X, y) can be O and 1, there may be only 0 and 255, may be any two values ​​of different compositions, the following formula:
Figure CN102096812AC00025
肤色检测,目前常见肤色检测检测中采用了各种各样的颜色空间,有RGB、HSV、YCbCr 与Lab等,肤色检测方法有基于非参数模型、基于参数模型、基于支持向量机(SVM)模型、基于神经网络模型、基于红外感知的肤色检测、基于近红外图像和基于高斯分布等方法。 Color detection, the current common skin color detection assay using a variety of color space, RGB, HSV, YCbCr and Lab etc., skin color detection methods are based on non-parametric model, based on the parameters of the model, based on support vector machine (SVM) model neural network model based on infrared detection based on skin color perception, based on near-infrared images and methods based on Gaussian distribution. 以上方法适用于本发明但不局限于以上方法,只要用到了人体肤色检测,均适用于本步骤。 The method of the present invention is more suitable for, but not limited to the above method, as long as the human skin used in the detection, are applicable to this step. 令Is(x,y)为肤色图像,Ι(χ,y)为红外图像,则肤色检测公式如下: Order Is (x, y) is a color image, Ι (χ, y) is an infrared image, the color detection formula is as follows:
Figure CN102096812AC00031
基于高斯分布的肤色检测方法步骤是,在YCbCr空间,对不同图像进行肤色采样,计算样本的均值M和协方差C,令P表示概率值,则肤色检测为肤色差分,表示使用当前帧肤色图减去背景帧肤色图,但不局限于使用当前帧,也可以使用当前时刻前后相邻若干时刻帧的图像,只要使用了背景帧肤色做减法,均在肤色差分范畴。 Based on the Gaussian distribution of skin color detection step, in the YCbCr space, different image color samples, calculate the sample mean M and covariance C, so that P denotes the probability value, the skin color detection is color difference, using the current frame complexion FIG subtracting the background frame color chart, but is not limited to the use of the current frame, you can use the time before and after the current time frame of several adjacent images, just use the background color frame subtraction, all in color difference category. 令Ibs(χ,y)代表背景帧肤色图像,Ics(χ,y)表示当前帧肤色图像,Ids(χ,y)表示肤色差分图像,差分公式如下: Order Ibs (χ, y) representative of the background color image frames, Ics (χ, y) represents the color image of the current frame, Ids (χ, y) represents color difference image, difference formula as follows:
Figure CN102096812AC00032
肤色运动检测与运动检测步骤相同;运动检测合并:即= /⑩,力,①表示二进制与,I' bin为帧间差分检测结果,Ibin为背景差分运动检测结果,并存放合并后检测结果。 Color motion detection and motion detection step of the same; combined motion detection: ie = / ⑩, force, and represents a binary ①, I 'bin for the interframe difference detection result, Ibin background differential motion detection result and the detection result of the combined store.
CN2011100323381A 2011-01-30 2011-01-30 Teacher blackboard writing action detection method for intelligent teaching recording and playing system CN102096812A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2011100323381A CN102096812A (en) 2011-01-30 2011-01-30 Teacher blackboard writing action detection method for intelligent teaching recording and playing system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2011100323381A CN102096812A (en) 2011-01-30 2011-01-30 Teacher blackboard writing action detection method for intelligent teaching recording and playing system

Publications (1)

Publication Number Publication Date
CN102096812A true CN102096812A (en) 2011-06-15

Family

ID=44129900

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2011100323381A CN102096812A (en) 2011-01-30 2011-01-30 Teacher blackboard writing action detection method for intelligent teaching recording and playing system

Country Status (1)

Country Link
CN (1) CN102096812A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104167116A (en) * 2014-07-15 2014-11-26 苏州阔地网络科技有限公司 Multi-moving object picture switching control method and system
CN104539888A (en) * 2014-12-16 2015-04-22 广西科技大学 Video monitoring method for closed-chest cardiac massage in cardio-pulmonary resuscitation first aid training
CN104618650A (en) * 2015-01-30 2015-05-13 苏州阔地网络科技有限公司 Movement detection method and device through combination between front image acquisition device and side image acquisition device

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2005135014A (en) * 2003-10-28 2005-05-26 Hitachi Kokusai Electric Inc Object detection system
CN101635833A (en) * 2008-07-22 2010-01-27 深圳市朗驰欣创科技有限公司 Method, device and system for video monitoring
CN101814242A (en) * 2010-04-13 2010-08-25 天津师范大学 Moving object real-time tracking recording device of classes

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2005135014A (en) * 2003-10-28 2005-05-26 Hitachi Kokusai Electric Inc Object detection system
CN101635833A (en) * 2008-07-22 2010-01-27 深圳市朗驰欣创科技有限公司 Method, device and system for video monitoring
CN101814242A (en) * 2010-04-13 2010-08-25 天津师范大学 Moving object real-time tracking recording device of classes

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104167116A (en) * 2014-07-15 2014-11-26 苏州阔地网络科技有限公司 Multi-moving object picture switching control method and system
CN104539888A (en) * 2014-12-16 2015-04-22 广西科技大学 Video monitoring method for closed-chest cardiac massage in cardio-pulmonary resuscitation first aid training
CN104539888B (en) * 2014-12-16 2018-06-05 广西科技大学 The video frequency monitoring method of closed cardiac massage art in CPR first aid training
CN104618650A (en) * 2015-01-30 2015-05-13 苏州阔地网络科技有限公司 Movement detection method and device through combination between front image acquisition device and side image acquisition device

Similar Documents

Publication Publication Date Title
Brdiczka et al. Learning situation models in a smart home
Fisher et al. Speaker association with signal-level audiovisual fusion
Fritsch et al. Audiovisual person tracking with a mobile robot
Orchard et al. Converting static image datasets to spiking neuromorphic datasets using saccades
TW201025193A (en) Method for automatic detection and tracking of multiple targets with multiple cameras and system therefor
CN104813339A (en) Methods, devices and systems for detecting objects in a video
Mathe et al. Actions in the eye: Dynamic gaze datasets and learnt saliency models for visual recognition
WO2004086301A3 (en) System and method for vehicle detection and tracking
Voit et al. Neural network-based head pose estimation and multi-view fusion
CN101719216A (en) Movement human abnormal behavior identification method based on template matching
CN104575142B (en) Seamless across the Media open teaching experiment room of experience type digitlization multi-screen
CN103093654B (en) Double video camera interactive intelligent tracking teaching system
CN105160310A (en) 3D (three-dimensional) convolutional neural network based human body behavior recognition method
WO2008098235A3 (en) System and method for providing simulated images through cosmetic monitoring
CN102411854B (en) Classroom teaching mixing technology application system based on enhanced reality and method thereof
CN101980248A (en) Improved visual attention model-based method of natural scene object detection
CN104200237A (en) High speed automatic multi-target tracking method based on coring relevant filtering
CN102053563A (en) Flight training data acquisition and quality evaluation system of analog machine
CN100357988C (en) Identification method of mobile number plate based on three-channel parallel artificial nerve network
US8611587B2 (en) Device, system and method for determining compliance with an instruction by a figure in an image
Rolf et al. Attention via synchrony: Making use of multimodal cues in social learning
CN103413120B (en) Tracking based on object globality and locality identification
TW201120408A (en) Object tracking method and apparatus for a non-overlapping-sensor network
CN104063883B (en) A kind of monitor video abstraction generating method being combined based on object and key frame
Ba et al. A study on visual focus of attention recognition from head pose in a meeting room

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
C02 Deemed withdrawal of patent application after publication (patent law 2001)