Research On Intelligent Monitoring Method Based On Examination Surveillance Video

With the promotion of standardized examination rooms' construction,shortcomings such as low efficiency of traditional video surveillance technology and massive video storage pressure have gradually highlighted.Intelligent monitoring system is an application of intelligent behavior analysis technology,which can put an end to lax supervision,improve monitoring efficiency and ease the pressure of massive video storage.Therefore,the research of intelligent monitoring method based on the examination surveillance video is of great practical significance,which not only reduces the input of human,financial and material resources,but also improves the fairness of examinations.In this paper,a set of intelligent monitoring method is put forward under the examination environment,which is applicable for the attendance statistics of examinees and the intelligent detection of the examinees'abnormal behavior.The main work of this paper is carried out in the following four aspects.(1)Through the observation of the characteristics of the examinees sitting,we propose an object detection method based on the part of head and shoulder.We combine Histograms of Oriented Gradients feature and Uniform Local Binary Pattern Histograms feature to form a new feature.Support Vector Machine was used to train classifiers respectively with the single feature and combined feature.The experiment was carried out on the examinees' experimental dataset and the detection performance is analyzed.We propose an examinee detection framework based on classifier cascade to meet the requirements of detection rate and detection speed.(2)Considering the particularity of the examination room environment,an error processing method which combines skin detection and hair detection based on YCbCr color space and the algorithm of RANdom SAmpling Consensus is put forward to correct the detection results to achieve the purpose of counting examinees and recording attendance.(3)This paper puts forward a method of detecting abnormal behaviors of examinees based on sparse reconstruction,which extracts spatio-temporal gradient feature to describe appearance characteristics of examinees' behaviors.The original sample data is simplified by extracting the moving saliency area and the Principal Component Analysis to reduce the calculation.The sample data of examinees' normal behaviors is used to do sparse combination learning and the model is established.Then we compute the corresponding reconstruction error for each test sample through the model to complete the abnormal behavior detection.On the experimental dataset in this paper,the method can achieve a good detection performance and a real-time detection speed.(4)Aiming at improving the performance of abnormal behavior detection of examinees based on sparse reconstruction,a method of motion blobs based on the motion history image is added to form a dual channel detection framework based on multi-information fusion.Under the dual channel,the performance of examinees'abnormal behavior detection has been effectively improved.By comparing with other common methods of examinees' suspicious behavior detection,it is proved that the method proposed in this paper has better universality.