A Video Stitching Method Based on Adaptive Key Frame Sampling

Abstract

The present invention discloses a video mosaic method based on adaptive key frame sampling. Aiming at the problems existing in the prior art, it revolves around how to utilize the space and time information of video frames, and performs image registration between video frames in a pre-estimated ROI, By estimating the coincidence rate, determining the key frame, adjusting the sampling step, etc., the key frame can be effectively extracted and the error matching can be reduced. At the same time, the sudden change in the video frame sequence can be effectively processed to achieve a more accurate and fast video splicing. At the same time, the key frame sampling interval is automatically adjusted, and the image registration is performed on the pre-estimated ROI to improve the efficiency and accuracy of image stitching. By reasonably extracting key frames, it also has good applicability under non-uniform changes in the video scene. The invention is simple and effective, and has good application prospects.

Description

technical field [0001] The invention belongs to the technical field of video splicing, and in particular, relates to a video splicing method based on adaptive key frame sampling. Background technique [0002] Image stitching is the basis of video stitching. Compared with static image stitching, video stitching requires real-time input video streams to be spliced ​​to obtain real-time output panoramic video streams or panoramas with wide continuous scenes. Video stitching has a wide range of applications, and has a wide range of applications in aerial video stitching, minimally invasive surgical equipment, video surveillance and other fields. [0003] Video stitching is to register the video frames with continuous scene changes at different moments of the same camera or with common scene parts acquired by different cameras, and seamlessly stitch them into a panoramic image with a larger field of view or a dynamic larger field of view video stream. Generally, video is regard...

AI Technical Summary

Problems solved by technology

[0006] Among them, literature [Liu Yong, Wang Guijin, Yao Anbang, et al. Video stitching based on adaptive frame sampling [J]. Journal of Tsinghua University: Natural Science Edition, 2010(1): 108-112.5] and Chu Ling, Guo Sanhua, Xie Shaoxia. A video sequence splicing method based on adaptive key frames [J]. Electronic Design Engineering, 2011, 19(21): 189-192.] The setting of key frames is fixed length, and can not be fully utilized The spatial and temporal correlation between images is only treated as a uniform motion for video motion, and the efficiency of key frame extraction for non-uniformly changing video frame sequences is not good, and a large number of false matches will be performed. Emerging mutations cannot be efficiently processed

Images