Research of algorithm for detecting suspected cheating behavior in examination surveillance video

摘要

It is difficult to review vast amounts of examination surveillance videos recently. In this paper, an algorithm for detecting suspected cheating behavior in examination surveillance video is proposed to solve this problem. The algorithm is based on adjacent frame difference algorithm, optical flow method and background difference method and optimizes them to detect suspected cheating from original examination surveillance videos. Then, the censors review the detected videos which contain the suspected cheating behavior and determine whether there is any cheating directly. This process is equivalent to review the original video review, thus the workload of reviewing original examination surveillance videos is reduced.