Intelligent video identifying method for cheat in test

Abstract

The embodiment of the invention provides an intelligent video identifying method for cheat in a test. The method comprises the following steps of: firstly, dividing all pixel points in an examination room monitoring video into foreground pixel points and background pixel points through establishing and updating a (GMM) Gaussian mixture model aiming at the examination room monitoring video; extracting sports pixel characteristic values according to foreground images and extracting the characteristic value of a testee area in the examination room as the judging data of a decision tree in the next step; and training the decision tree by using a series of sample legends to obtain a preset threshold value of each node and probability values of final layer nodes corresponding to different cheating accidents and judging the corresponding cheating accidents according to the data of the decision tree. Through the technical scheme, the cheating behaviors and nonfeasance phenomena of supervisorsin the examination room monitoring video can be automatically detected, and staff is freed from staring the video constantly to search cheating accidents, therefore labor, materials and financial resources are saved greatly, and the working efficiency is improved.

Description

technical field [0001] The invention relates to the field of video monitoring, in particular to a method for intelligent video recognition of cheating events in exams. Background technique [0002] At present, with the gradual deepening of computer vision technology and artificial intelligence theory research and the further promotion of its application, the monitoring system is directed to make more use of video image understanding and intelligent information processing technology, making the system appear intelligent and forming intelligent video surveillance system. The event-based intelligent video analysis technology can use the powerful data processing function of the computer to analyze the massive data in the video screen at high speed, extract the events that the user cares about, and provide useful key information for the monitor. [0003] Event-based video surveillance can be said to be a relatively new field in my country, but there are already many units concer...

AI Technical Summary

Problems solved by technology

[0004] In the field of educational examinations, in order to create a good, fair and fair national education unified examination, it is often necessary to use intelligent monitoring technology to monitor violations. However, it is unrealistic to detect fraud by completely relying on human eyes to watch videos Unscientific. For example, in 2007, Liaoning Province used about 200 teachers to review nearly 100,000 hours of college entrance examination video materials one by one in order to verify cheating, which took a lot of time and labor.

[0005] It can be seen that in the prior art, although the electronic monitoring system uses cameras to monitor the examination room on-site, it needs to manually judge and process various events in the examination room, thus consuming a lot of manpower and material resources.

Images