Multi-student standing and sitting detection method based on optical flow and k-means clustering

A detection method and a student's technology, applied in the field of information teaching, can solve the problems of complex installation and wiring, not beautiful, high cost, etc., and achieve the effect of high detection accuracy, good real-time performance, and simple equipment

Inactive Publication Date: 2017-05-31
TIANJIN UNIV
4 Cites 12 Cited by

AI-Extracted Technical Summary

Problems solved by technology

Although this can finally achieve the purpose of detection, but the cost is high, the installation and wiring are complicated, and the camera needs to be installed on one side of the classroom, which is close to the height of the ...
View more

Abstract

The invention discloses a multi-student standing and sitting detection method based on an optical flow and k-means clustering, and the method comprises the steps: collecting image information of a student region; carrying out the moving object region detection for the student region, detecting a moving region in an image through an inter-frame difference method, and marking a moving object region template through a connection domain marking algorithm; enabling the moving object region template to be superposed with an original image, carrying out the optical flow calculation of the moving object region, and obtaining a motion vector of a feature point; enabling the optical flow to be divided into an upward group A and a downward B, and carrying out the clustering; determining that the object performs standing and sitting motions if the number of continuous appearing times of the same central point reaches a certain value, and ending the detection at this moment. Compared with the prior art, the method can achieve the detection of the standing and sitting behaviors of the students through one camera, and the equipment is simple and easy to install. The method is suitable for various types of environments. The method is high in detection accuracy, is better in instantaneity, and can achieve the class demands. The method can achieve the detection of simultaneous standing condition of a plurality of students, and is more practical.

Application Domain

Image enhancementImage analysis +1

Technology Topic

Optical flowRegion detection +4

Image

  • Multi-student standing and sitting detection method based on optical flow and k-means clustering
  • Multi-student standing and sitting detection method based on optical flow and k-means clustering
  • Multi-student standing and sitting detection method based on optical flow and k-means clustering

Examples

  • Experimental program(1)

Example Embodiment

[0020] The embodiments of the present invention will be described in detail below in conjunction with the drawings.
[0021] Such as figure 1 As shown, a schematic diagram of the installation of a camera in an embodiment of the present invention. The camera is installed above the wall in front of the student area in the classroom, and the position is not limited; you can also choose a suitable place to install it.
[0022] Such as figure 2 As shown, it is a flowchart of the multi-student sit-up detection method based on optical flow and k-means clustering of the present invention, which specifically includes the following steps:
[0023] Step 1. Video capture: enable the camera, collect the video information of the student area, and send it to the analysis and processing system;
[0024] Step 2. Detect the motion area: detect the motion area of ​​the student area, and use the inter-frame difference method to detect the motion area in the image: calculate the grayscale difference of each point between the two frames before and after, and use the absolute value of the difference As the gray value of the image after the difference. Perform threshold segmentation on this image to get a binary image. The schematic diagram of the result is shown in Figure (3a);
[0025] Then, use morphological processing such as expansion and corrosion to connect the areas to remove noise, and use the connected domain labeling algorithm to mark the moving area template: perform the closed operation of first expansion and then corrosion on the picture (3a) to remove the isolated noise and make the white area more Unicom, the rendering is shown in Figure (3b), and then the connected domain labeling algorithm is used to obtain the motion area template shown in Figure (3c);
[0026] Step 3. Find the optical flow field of the motion area: superimpose the motion area template with the original image to obtain the original image information of the motion area, and perform optical flow calculations on the motion area to obtain the motion vector of the feature point in the motion area: first use The Shi-Tomas algorithm finds the corner points in the motion area, and then uses the LK optical flow method to calculate the optical flow of each corner point, that is, the motion vector. The result is shown in Figure (3d).
[0027] Step 4. Divide the optical flow into two groups: If the vertical component of the motion vector> The absolute value of 5px and the horizontal component is not greater than 3pixel, it is considered that the feature point has an upward movement trend, and these feature points are grouped into a group, denoted as A, if the vertical component of the motion vector
[0028] Perform k-means clustering on group A, you can divide all points in group A into K categories, and get the center position coordinates of each category. At this time, the number of categories is the number of students who tend to stand up in the current frame , The center position of each class is the center coordinate of each target; the same k-means clustering of group B can divide all points in group B into K classes, and get the center position coordinates of each class, At this time, the number of classes is the number of students who tend to sit down in the current frame, and the center position of each class is the center coordinates of each target area;
[0029] If the number of samples in a class is too small, such as when the number of samples is less than 5, the class is considered to be an interference point, this class is excluded, and the center coordinates of other classes are recorded;
[0030] Go back to the first step and perform the same processing on the next frame of video;
[0031] If the same center coordinate in group A (the x coordinate is approximately unchanged, the y coordinate continues to decrease and the difference does not exceed 10px each time is considered to be the center coordinate of the same target) 20 consecutive times, it is considered that the target has a standing action. In group B, the same center coordinate (x coordinate is approximately unchanged, y coordinate continues to increase and the difference does not exceed 15 pixels each time is considered to be the center coordinate of the same target) 15 consecutive times, it is considered that the target has a sitting action. At this point, the test is over.

PUM

no PUM

Description & Claims & Application Information

We can also present the details of the Description, Claims and Application information to help users get a comprehensive understanding of the technical details of the patent, such as background art, summary of invention, brief description of drawings, description of embodiments, and other original content. On the other hand, users can also determine the specific scope of protection of the technology through the list of claims; as well as understand the changes in the life cycle of the technology with the presentation of the patent timeline. Login to view more.

Similar technology patents

Process for preparing notoginsen triterpenes

InactiveCN1634970Aincrease the rate of submissionsimple equipment
Owner:张平

Video anti-shake method based on feature point matching

InactiveCN103813099Asimple equipmentlow cost
Owner:RES INST OF SUN YAT SEN UNIV & SHENZHEN

Classification and recommendation of technical efficacy words

  • simple equipment
  • easy installation

Innocent treatment of cyanide and fluoride in solid waste residues and recycling process thereof

ActiveCN101444660Asimple equipmentreduce land use
Owner:浙江钰滔环保科技有限公司

Telegraph pole installation robot

Owner:浙江邮电职业技术学院 +1

Quadruple frequency-based scheme for realizing base station passive full duplex millimeter wave RoF link

InactiveCN101742738Asimple equipmentStrong practical operability
Owner:BEIJING UNIV OF POSTS & TELECOMM

Core sample sealing device in true triaxial test

Owner:INST OF ROCK AND SOIL MECHANICS - CHINESE ACAD OF SCI

Automatic sorting and boxing equipment of component assembly line based on vision

Owner:HEFEI INSTITUTES OF PHYSICAL SCIENCE - CHINESE ACAD OF SCI

Apple bagging mechanism

Owner:SHENYANG INST OF AUTOMATION - CHINESE ACAD OF SCI

Vehicle-mounted hoisting fixed platform of rotor-wing type unmanned aerial vehicle

ActiveCN106904288Aeasy installationprone to water accumulation
Owner:BEIJING INSTITUTE OF TECHNOLOGYGY
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products