Embedded panoramic video splicing system based on TX2

Abstract

The invention discloses an embedded panoramic video splicing system based on TX2. The system can be used for observing an external battlefield environment, expanding a visual field, identifying and judging surrounding environment threats and abnormities, and comprehensively improving the battlefield environment sensing capability of combat personnel. In a rapid driving process of war vehicles such as tanks, vehicle passengers in the tank can observe 360-degree information around the vehicle in real time by using the tank. The military sensitive target in the video is automatically detected and recognized, the military target information of the battlefield environment outside the vehicle is mastered in real time, the head or body does not need to be stretched out of the vehicle, the observation view field is expanded, the reconnaissance effect is improved, and the personal safety of combat personnel can be guaranteed.

Description

technical field [0001] The invention belongs to the fields of image / video processing and artificial intelligence, and in particular relates to an embedded panoramic video splicing system based on TX2. Background technique [0002] Image stitching is the core and foundation of video stitching, and the problem of image stitching has a long history. In real life, it is often desired to obtain a large-scale scene, but due to the limited shooting angle of the camera, it is difficult to capture all the expected information. Although some professional cameras, wide-angle cameras and fisheye cameras have a relatively wide viewing angle, they are expensive on the one hand and have serious image distortion on the other, which still cannot meet the demand. Therefore, if you want to align multiple small-field images with overlapping areas to form a large-field high-resolution image, this is image stitching technology. Image stitching effectively solves the problem of low resolution of...

AI Technical Summary

Problems solved by technology

However, the current video sources used for detection and recognition are basically collected when the camera is still and controllable. In a complex combat environment, how to accurately and quickly target the key targets in the video for the video collected when the vehicle is moving rapidly and bumpy? There is no relevant research on detection and identification

Images