Vsurv: A Visualization Scheme for UAV Reconnaissance Video

Abstract

This paper proposes a visualization scheme for UAV reconnaissance video to describe the video content and structure in a user-friendly way. Firstly, the video content is analyzed efficiently by the feature of Image perceptual hashing moment proposed to measure the content importance of frame, secondly, the bag-of-words model is applied to extract keyframes' semantic information from the entire video, finally, the video description is generated through visualization model. The experiment result shows that visualization scheme can display the video's content, structure and highlight clearly to users.