A Video Compensation Method for Eliminating Line Segment Intersection Mis-match Points

Abstract

The invention provides a video compensation method for eliminating line segment crossing mismatch points, the video compensation method includes the following steps: A, video frame input; B, feature point extraction in the current frame and a reference frame, wherein the reference frame is the current The image of the previous frame of the frame; C, match the feature points of the reference frame and the current frame image; D, eliminate the matching point pairs with line segment intersections; E, randomly select three feature points located on the reference frame, and correspond to them in the current frame The matching points of the three pairs of matching points are used to obtain the transformation matrix; F, for each point in the current frame image, using the transformation matrix to obtain the compensated image after image interpolation. The invention utilizes the feature that the connection between the video frames is relatively close, eliminates the wrong matching points where the line segments intersect, improves the ratio of the correct matching point pairs, and makes the calculation of the affine transformation matrix more accurate.

Description

technical field [0001] The invention relates to a video compensation method for eliminating line segment crossing mismatch points. Background technique [0002] Camera systems used in vehicles, UAVs, ships and other moving platforms or field monitoring systems often cause random jitter in the output image sequence due to their own motion, which adversely affects video monitoring and subsequent image processing. Eliminate this image jitter The latest technology is called image stabilization technology, and image stabilization technology has gradually transitioned from the initial mechanical image stabilization and optical image stabilization to the electronic image stabilization (Electronic Image Stabilization) stage. The video frame compensation link of electronic image stabilization generally needs to establish an affine transformation model, and the calculation of the affine transformation matrix needs to use the image matching points of the reference frame and the current...

AI Technical Summary

Problems solved by technology

[0005] When the above technical solution uses the image feature point matching method based on SIFT and SURF to calculate the video frame affine transformation matrix, only the size of the feature point feature value is used for feature point matching, and there is no specificity for motion between video frames-adjacent The rotation and translation movement between videos is small, and the motion displacement vectors between matching feature points are similar, resulting in mismatching points with line segment crossings, and such mismatching points bring errors to the calculation of the affine transformation matrix

Images