Reseach And Application Of Breathing Detection Based On Eulerian Video Magnification Algorithm

With the rapid development of society,the status of the improvement of the material level and large-scale population movement in China,security awareness has gradually been put on the agenda.As a powerful means of modern security,video surveillance has been paid more and more attention.Video surveillance system has also been widely used,and video intelligent behavior analysis has been paid more and more attention.Focus on the monitoring of special circumstances such as prisons,every movement of the key monitoring person at any time should be in sight.Especially the key monitoring person cannot tolerate the current environment,pressure or such an endless day,they may even have the urge to commit suicide.When the key monitoring person are sick,they lie in bed to have a rest,especially in the quilt,ordinary video monitoring cannot judge whether the person is under normal rest,so the breath detection is particularly important.In this paper,combined with the existing research status,a breath detection method based on video analysis is designed,which meets the requirements of accuracy and real-time.The main work of this paper is as follows:1.Considering the relative stability and simplicity of the indoor environment,the detection of foreground objects based on motion history image is proposed.Firstly,motion is transformed into time function by motion history image.Then the time function image is transformed into grayscale image,and the foreground moving target is obtained by thresholding.Then all the moving objects are obtained by the analysis of connected domain and motion segmentation.The motion history image can detect the foreground object and can also effectively alleviate the "empty hole" phenomenon of method named frame to frame difference.At the same time,the frame to frame difference method is simple to calculate and can satisfy the real-time performance on the basis of satisfying the system's accuracy.Therefore,the algorithm proposed in this paper can meet the requirements of the accuracy and real-time of the system based on the motion history image.2.A tracking algorithm based on region and pyramid L-K tracking algorithm is proposed.The initial position of the tracking target is obtained by the region-based tracking algorithm,and then the initial position of the target is transferred to the pyramid L-K tracking algorithm to track the target trajectory.In this paper,the pyramid L-K tracking algorithm has high precision,which can meet the precision of indoor tracking.3.A new breathing detection algorithm based on Eulerian Video Magnification algorithm combined with grid marking and motion history image is proposed.Firstly,the original image is marked with double-layer grids,and then get movement through motion history image and combined with grid markers to remove the movement of other than breathing interference.Then,the small movement of breathing is enlarged by Eulerian Video Magnification algorithm.The next step is to compute the cycle of foreground points to obtain the cycle of breathing and judge whether the breathing is normal.In this paper,the breathing detection every four frames processed once,and the sampling area down sampling,to meet the real-time system requirements.Experimental results show that the detection and tracking of foreground objects and the Euler image-based algorithm are feasible and can meet the requirement of accuracy and real-time.