Detection And Analysis Of Abnormal Behavior In Electronic Invigilating

At present, as a video monitor development direction, the electronic monitor has been widely applied to various schools, all kinds of tests. With the popularity of electronic monitors in the college entrance examination, this year, the electronic supervision has been applied to the national postgraduate entrance exam for the first time this year, so it can foresee that electronic monitor will play an increasing important role on test order and the emergencies during examination. But the current electronic invigilation still uses the traditional monitoring methods, the workers need monitor the whole screen so that it may cause the visual fatigue and miss the abnormal behavior in time, but also large amounts of video datas need to be stored aferwards, which is inconvenient to research. The application of intelligent video monitoring in the examination room, can make electronic monitors get rid of the dependence of the traditional monitoring to human resources, so finally it realizes the Intelligent Electronic monitors (Intelligent Electronic Invigilation, IVI).Based on the key technology of human behavior of video sequences, research and analysis the identification technology of abnormal behavior, around the examination room. A variety of abnormal behavior modeling of electronic monitors, video segmentation, template matching and other key technologies are studied, in order to achieve the examination of abnormal behavior that can be warned, make report to the police, and after retrieval under the background of examination room. This particular for accurate identification of abnormal behavior.The mainly researching content and results of this article are as follows: (1) Studied of the existing modeling methods, including Gaussian Mixture Model, Feature Point Matching, the human body skeleton model and the background model. To test the common abnormal behavior for accurate definitions and set up determine standard, and to modeling a variety of abnormal behavior, achieving them by different methods. Through experiments show that the model based on a variety of abnormal behavior can be more quickly and accurate recognizing the abnormal behavior.(2) Studied the technology of video segmentation. Including video segmentation based on time, video segmentation algorithm based on space technology and merging the space and time of video segmentation technology, this paper expounds the semiautomatic video segmentation technology principle, in the examination of many background. Through the difference background methods, segmenting the region that contains a single examinee’s sequence, located on a single target, through the semi-automatic segmentation technology to segment each target that using the integration method in time and space. Experiments show that this testing method is conforming the specific background, and reducing the amount of calculation, and improving the efficiency of segmentation.(3) Based on the common abnormal behavior modeling and the video region contour segmentation, using the feature template matching for examination of abnormal behavior detection, this algorithm is able to detect abnormal behavior of the examination. Through the introduction of principle of the template matching method, the contour of Canny operator can be used to detect the image, selecting HMM model is trained on the abnormal behavior, for improving the testing efficiency, then recognizing two common abnormal behaviors in the examination room based on template matching. Experiment shows that compared with other methods, the methods of template matching can be quickly and accurate recognizing in the specific examination background.