Research And Implementation Of Multi-target Identification And Tracking Algorithm In The360-degree Full-scene

In recent years, with the society progress and the continuous development of science andtechnology, more and more occasions need for real-time video monitoring, especially Banks,government departments, even border and other sensitive areas, but also for video monitoringdemand is higher and higher, more and more accurate. Previous video intelligent monitoring needsecurity personnel to real-time monitor each area, although with lots of pictures, long time andall-weather monitoring, it’s hard for simple humans, and most of the camera is fixed Anglemonitoring.There are many panoramic monitoring cameras, and this type of camera generally is fisheyelens, but the price is relatively jigh, there are also panoramic cameras use several normal lensjoining together. Panoramic camera although can cover a wide range of area, and even can achieve360degree panoramic no dead Angle monitoring. Because of the fisheye lens,the focal length isvery short, and the monitoring image distortion is relatively serious.When the distances exceed fiveto ten meters, it is difficult to see faces. For countries such as the border to a wide range ofmonitoring area, this kind of panoramic camera is difficult to meet the requirements.This paper trys to research that360degree panoramic motion object detection and real-timetracking based on image splicing. This paper represents image splicing algorithm and the trackingand multi-target recognition algorithm base on splicing technology. In this paper the advantages ofthe algorithm is common rotating camera use software technology still can achieve360degreepanoramic monitoring, it reduced the cost, and the monitoring distance will not be affected. Firstly,this paper researched the traditional splicing algorithm,according to the moving objects detection ofreal-time and accuracy requirements, using SIFT algorithm for video image sequence to takepanoramic joining together, in order to later real-time motion object detection. Secondly,this paperresearched motion object detection and tracking based on the panoramic image splicing algorithm.Finally,the system achieved the expected design goal through the system test.