Real time video image stabilization method based on simple fast global motion parameter estimation

Abstract

The invention discloses a real time video image stabilization method based on simple fast global motion parameter estimation, which is simple to realize, has higher estimation prevision and provides a very effective method for real time video image stabilization under the environment of a low-speed embedded type platform. The real time video image stabilization method based on simple fast global motion parameter estimation comprises the following steps of: 1, carrying out block partition on a current frame; 2, obtaining a motion vector of each block based on a diamond search algorithm; 3, judging whether the motion is abnormal, and marking the block to be an effective block if the motion is not abnormal; 4, counting an average value of the translation invariants of all effective blocks to estimate a global translation invariant; 5, estimating rotation and zoom quantity of each effective block relative to a match block in a previous frame according to the global translation invariant; 6, counting all rotation and zoom qualities to estimate the global rotation and zoom quantity; 7, carrying out the motion filter process on the global translation invariant and the global rotation and zoom quantity; and 8, carrying out the motion compensation according to filter results.

Description

technical field [0001] The invention relates to a real-time video stabilization system that can be widely used in consumer electronics products such as mobile phones, digital cameras, and video cameras, and can also be widely used in camera and monitoring systems equipped with aircraft, ships, satellites, and missiles. method. Background technique [0002] The video sequence shot by the camera during the movement not only includes the active part of the camera, but also introduces irregular random motion. The existence of this random motion will cause the video to shake and blur, which seriously affects human vision. Feeling, the purpose of video image stabilization is to eliminate or weaken the destructive effects of random motion on video images. The realization of video image stabilization includes three modules: motion estimation, motion filtering and invalid area compensation. For these three modules, domestic and foreign scholars have proposed many methods. In terms ...

AI Technical Summary

Problems solved by technology

[0004] Finally, after searching the prior art documents, it was found that Marius Tico published "Constraint motion filtering for videostabilization" (video stabilization method based on constrained motion filtering, image processing IEEE International Conference), this paper adopts the feature point tracking method and the Kalman motion filter method with constraints. , the precision is high, but the speed is slow. At the same time, the Kalman motion filtering method based on the uniform velocity model is adopted in this paper. When the motion model matches, the filtering effect is better, but once the model changes, the filtering effect will be greatly affected. The camera often does not conform to the law of uniform motion during the shooting process

Images