基于时间配准的视频图像稳定方法

Research of a Video Stabilization Algorithm Based on Time Registration

在车载视频监控系统中,由于载体的姿态变化或振动的影响,使得获得的图像信息模糊、不稳定,增加了观察者的疲劳度。为了解决车辆行进间视场抖动、图像模糊的问题,需要对实时视频图像进行稳定处理,将视频采集装置直接固定在初步稳定的测量平台上,通过稳定平台的稳定残差进行二次图像稳定,但由于传输延迟的存在,对接收到的传感器数据和视频采集的图像数据分别进行了队列缓存,即加入了时间配准方法。通过在某遥控车上验证,其能保证实时视频图像获得最佳的稳定效果。

In a vehicle Video Monitor System,the acquired images are vague and unstable because of the varied postures and vibrations of the vehicle while it is moving.The vibrations of the field of view （FOV） and vague images increase the fatigue of observers.As a result,we need to stabilize the real-time images to solve this problem.The proposal of this paper is to fix the video capture device directly on the preliminary stable measurement platform,whose stable residuals are applied to second-time stabilization.However,due to the delay during the transfer,the data received from the censors and images acquired by the video will be queue buffered respectively,wbich is called time registration.After testing on a remote control vehicle,this proposal has been verified that the acquired real-time video images could achieve optimal stabilized result.