Font Size: a A A

Research On Surveillance Video Synopsis Based On Spatio-temporal Tube

Posted on:2024-08-10Degree:MasterType:Thesis
Country:ChinaCandidate:K N GuoFull Text:PDF
GTID:2568307151953339Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
With the enhancement of people’s security awareness,video surveillance system has been widely used,resulting in a growing amount of surveillance video data.How to quickly obtain the required content from a large number of surveillance video data has an urgent practical demand.The video synopsis technology expresses the content of the original surveillance video with short condensed video.In order to solve the problems of high computational complexity and serious pseudo collision of video concentration,this paper studies the surveillance video synopsis method based on spatio-temporal tube.Firstly,the preliminary synopsis based on segments is carried out on the surveillance video,namely motion segment extraction.The static segments are deleted to completely retain the spatio-temporal information of the object.Then,video synopsis based on objects are carried out to exchange space for time and further enrich the video to improve the compression rate.The innovative achievements and main work completed are as follows:(1)A fast motion segment extraction method via nested ellipse spatio-temporal tubes in surveillance video is proposed.Aiming at the problem that most of the existing motion segments extraction methods are complicated and computative,which cannot meet the practical demand of fast extraction of moving segments.A fast motion segment extraction method via nested ellipse spatio-temporal tubes in surveillance video to minimize the amount of computation.Firstly,surveillance video is elliptically spatio-temporal sampled to form an elliptical spatio-temporal tube.Secondly,multiple elliptical spatio-temporal tubes sampled progressively according to surveillance scene are integrated to nested elliptical spatio-temporal tubes.Then,nested elliptical spatio-temporal tubes are expanded to generate spatio-temporal plane maps.Finally,the background of spatio-temporal plane maps is removed and the spatio-temporal flow model is constructed to extract motion segments.Experimental results show that the proposed algorithm has obvious advantage in calculating time,greatly reduces the amount of calculation under the premise of ensuring detection accuracy,and it can realize fast motion segments extraction in surveillance videos.(2)An adaptive motion segments extraction of surveillance video by combining dichotomous and spatio-temporal tube is proposed:Aiming at the problems that the existing motion segment extraction methods cannot take into account the computing speed,accuracy and poor robustness of the surveillance video with variable motion conditions,a method called adaptive motion segments extraction of surveillance video by combining dichotomous and spatio-temporal tube is proposed.Firstly,the initial spatio-temporal flow is calculated using nested elliptical spatio-temporal tube model to judge the completeness of target trajectory.Secondly,we adjust dynamically the elliptical sampling line by combining the bisection to capture adaptively the moving target in the sampling area.Finally,the pixels on the sampling line are extracted to form the adaptive spatio-temporal tube for motion segmentation.Experimental results demonstrate that the proposed method outperforms the state-of-the-art methods in terms of both computing speed and accuracy,which has strong robustness and is also suitable for surveillance scenarios with changing motion.(3)An efficient surveillance video synopsis method based on spatio-temporal rotation is proposed:Aiming at the problem that the video enrichment technology has serious collision artifacts and cannot accurately reflect the motion relationship between the original video objects,an efficient surveillance video synopsis based on spatio-temporal rotation is proposed.Firstly,an object interaction function is proposed to analyze the interactivity between objects,and the tube sets are divided for common processing.Secondly,the collision artifacts between the objects are analyzed.The dynamic time domain translation and spatio-temporal rotation are respectively carried out according to the object proportion threshold to avoid collision artifacts.And the angle critical threshold is defined to divide the spatio-temporal rotation into adaptive spatio-temporal rotation and critical spatio-temporal rotation,so as to avoid the collision artifacts while ensuring the compression rate.Thirdly,a chronological function is designed for timing judgment to avoid ambiguity caused by the confusion of disordered objects.Finally,the objects tubes are stitched onto the background to produce a synopsis video.The experimental results show that the frame condensation ratio of the proposed method is better than that of the comparison methods,and the collision artifacts can be avoided under the premise of guaranteeing the compression rate,which has good visual effect.
Keywords/Search Tags:video synopsis, motion segment extraction, spatio-temporal tube, spatio-temporal rotation, spatio-temporal slice
PDF Full Text Request
Related items