Fixed-point directional video real-time mosaic method without valid overlapping variable structure

Abstract

The invention discloses a fixed-point directional video real-time mosaic method without a valid overlapping variable structure. The method comprises the steps: respectively collecting the video stream information in different positions; dividing the compressed video stream information to multiple first static video frame groups according to the time sequence; converting the static image corresponding to each route of the video stream information in the first static video frame groups to the overlook picture of a shot object, and forming a second static video frame group; performing the location, full view coarse mosaic, compensation fusion, and full view fine mosaic to each video frame in the second static video frame group to obtain the real-time full view video stream. The method is capable of, in allusion to different perspectives and different directions, performing the mosaic processing to the image without valid overlapping variable structure which is collected in the same field, and generating the video stream in real time. The mosaic method is capable of improving the image mosaic precision, ensuring the image mosaic efficiency, and satisfying the real-time request of the video stream mosaic.

Description

technical field [0001] The invention relates to the technical field of real-time video image splicing integrated circuits, and more particularly relates to a fixed-point directional video real-time splicing device without an effective overlapping area. Background technique [0002] Video splicing devices are widely used in the fields of high-resolution images with large viewing angles, virtual reality, medical images, remote sensing technology, and military fields. As the image stitching problem becomes more and more complex, and the image stitching algorithm has higher and higher requirements for real-time and precision, the performance requirements of the device for realizing video stitching are also getting higher and higher. For example, for the general video stitching problem, the real-time stitching of video streams is generally achieved by transplanting its image stitching algorithm and real-time video synthesis algorithm to parallel computing architecture platforms s...

AI Technical Summary

Problems solved by technology

Obviously, this patent hopes to improve the stitching efficiency by simplifying the image stitching algorithm, but for a complex stitching problem, the simplification of the stitching algorithm will definitely lead to a decrease in the stitching quality

At the same time, the single serial graphic splicing device used in this patent cannot meet the real-time requirements of video stream splicing with large data volume and complex calculations, and its practicability is poor.

Images