Research On The Integration Of The Surveillance Video And The Two-dimensional Map

The real world is always in the state of dynamic changes,so how to thoroughly perceive and express this kind of dynamic space environment has become an urgent problem to solve presently.Traditional two-dimensional map is an important tool for people to express geospatial information with its advantages of macro,integrity and simplicity.However,the two-dimensional map is always giving priority to show static objects presently.Although there are multi-temporal image data,this production cycle is longer and it is difficult to reflect the dynamic changes of scene,which is unable to realize the real-time visualization of dynamic target and is more difficult to meet the needs of thorough expression and cognition for today's rapidly changing of real world.With the development of intelligent city,the number of video sensors increase substantially in the city,we can access surveillance video information easier.Not only can surveillance videos capture urban dynamic information in time,but it can also express a strong sense of reality of geographical space.Whereas,the monitoring mode of traditional storyboard is still the main monitoring mode of video monitoring system at present,the application limitations of which are mutually isolated of the pictures and lack of relevancy,and it's not conducive to grasp dynamic information of the whole region from the macro.Therefore,on the one hand the surveillance videos integrated with two-dimensional maps enhance two-dimensional static maps' ability to express the dynamic target,which makes the maps immersive.On the other hand it breaks through the limitations of the traditional video monitoring system to view separately in each sub-lens,which has achieved the efficient control for the overall situation in the big scene monitoring area.Based on the above research background and significance,this paper regards surveillance videos and two-dimensional maps as the research object,carrying out the integration of the surveillance videos and two-dimensional maps to realize the enhanced expression.The research contents and results are as follows:(1)This paper develops a geometric rectification method of large dip angle video image based on dynamic foreground separation.This paper will integrate surveillance videos from a large angle and two-dimensional maps of orthographic projection,so we will do geometric rectification for the surveillance video images.By separating the foreground and background of the videos,carrying out geometric rectification for the video static background and extracting dynamic targets of the foreground video.At last,the corrected video background image and the extracted foreground dynamic target will together mapped onto a two-dimensional map to achieve the integration of monitoring video and two-dimensional map.This method not only solves the dynamic target distortions after the geometric correction of a large dip angle image,but also can effectively improve the efficiency of geometric rectification of the monitoring video image.(2)Researching on the integration method for the overlapped area of surveillance videos and two-dimensional maps.Improving image mosaic algorithm based on SIFT feature points and splicing the overlap areas of surveillance videos' background images,this method improves the efficiency of image mosaic algorithm.This paper puts forward a strategy that video foreground object matching and making overlap areas trade-off to realize the integration of dynamic targets in the video overlapped areas and two-dimensional maps.The background and foreground of videos are processed separately,this method avoids such mismatches and splicing ghosting phenomena caused by foreground parallax in the direct splicing.(3)Developed a movement tendency deductive model that the dynamic target in a monitoring blind area based on online learning.This model is based on Bayesian network model and network constraints,which introduces the idea of online learning algorithm and makes Bayesian network parameters be dynamically modified in time with the changes in the environment,to realize intelligent estimations of dynamic target movement tendency in the video monitoring blind area,which forms a completely dynamic target flow in the two-dimensional maps,to realize the real-time dynamic perception of overall distributions of dynamic targets in the great geographical scenarios.(4)Developing a prototype system of integrating of the surveillance videos and two-dimensional maps.Based on the above research results,designing and developing a prototype system of integrating of the surveillance videos and two-dimensional maps.The prototype system includes the front-end display system for ordinary users and the backstage management system for administrators,integrating the videos of large dip angle for geometric rectification,video splicing and monitoring blind area deduction,which provides security domain a new perspective of observation and a new auxiliary decision-making method.