| Scenery shoot is shown intuitionistic with photoelectric system accord with observation customs of human eyes. The effect and quality of scenery is affiliated to with stability of platform. The more stable and vivid of camera with good platform, it is propitious to observation for operator. It is usual that photo electricity system on the vehicle, camera device by hand, video flat roof on the helicopter, plane and warship. With the complexity and variety of application environment, it is inevitably that the roof plat shakes out-of-order. But the shake will lead to the image become unstable; it is very difficult for human scout and image analysis deeply, so electric stable scenery technology is emerged.Electric stable image technology is a kind of method that to compensate smoothly the dynamic image sequence acquired by camera with random movement or dithering, the stable image sequence can be output smoothly. The dissertation studies the relative technology of stable image deeply on the basis of the foregone research, presents an electric stable image method on optical flow movement vector estimations. Firstly, it chooses the corner of object edge on the estimation graph, builds characteristic region, and realizes calculation of subsequent optical flow movement on the region. Secondly, in the process of calculation, it optimizes differential coefficient operator, meanwhile it uses pyramid multi-resolution policy to match initially on the low-resolution image and satisfy the hypothesis that the movement of optical flow is continuously, and the policy can save time cost of the saving space and it can be realized easily in engineering. Thirdly, it adapts the Kalman Filter to remain the active movement and separate accurate high frequency movement for image compensation. Finally, it designs the whole electric stable image arithmetic, and presents the evaluation quota, it analysis-the effect of the method by examples.The advantage of the methods in the article is described as follows:Firstly, high flexibility of traditional optical flow method is reduced consumedly; however, it can not reduce the estimation precision of high performance with strong reliability.Secondly, the method can obtain high frequent movement and remain active movement more precisely. It agrees in the observation rabbit of human eyes and benefit to the scout for human eyes. |