Video stabilization method, device and computer-readable storage medium

Abstract

The present application provides a video image stabilization method, device, and computer-readable storage medium, which relate to the technical fields of image processing and computer vision. According to the feature points of the first frame image in the N frames of images in the video, the method respectively obtains N-1 optical flow point pairs between two adjacent frames of images in the N frames of images that meet the optical flow point pair threshold; and then screens Out of N-1 optical flow point pairs, the candidate feature points that meet the first preset requirements and the background feature points and foreground feature points in each frame of the image corresponding to the candidate feature points are extracted, and each frame of the image in the video is extracted The feature points of the non-moving foreground object in the preset frame number image, and the feature points of the relatively static background object in the preset frame number image; when using this method to stabilize the video, when the foreground between video sequences When there are many moving objects or the movement is large, it has a better image stabilization effect, and has stronger robustness and accuracy.

Description

technical field [0001] The present application relates to the technical fields of image processing and computer vision, and in particular to a video image stabilization method, device and computer-readable storage medium. Background technique [0002] As a method of video image stabilization, electronic image stabilization is based on motion estimation between consecutive video images, and then performs motion filtering and motion compensation processing on each frame image in the video to obtain a stable image. Electronic image stabilization first performs global motion estimation on the image sequence, and after obtaining the motion parameters, conducts a comprehensive evaluation, then performs motion compensation, and finally obtains a stable output sequence. [0003] In the prior art, the optical flow method, as one of the motion estimation methods, is used to describe the motion of the observed object, surface or edge caused by the motion of the observer. The traditiona...

AI Technical Summary

Problems solved by technology

[0003] In the prior art, the optical flow method, as one of the motion estimation methods, is used to describe the motion of the observed object, surface or edge caused by the motion of the observer. The traditional optical flow method can be used to calculate the motion including translation, rotation And the transformation parameters of zooming motion, and have high precision, but when there are many foreground moving objects or large motions between video sequences, the global motion vector and transformation parameters calculated by the traditional optical flow method have large errors, which limits Its engineering application

Images