# VideoSynopsis视频摘要

**Repository Path**: jun-tian/video-synopsis-video-summary

## Basic Information

- **Project Name**: VideoSynopsis视频摘要
- **Description**: Web-based video synopsis system using Apache Spark。
基于Apache Spark的Web视频摘要系统
- **Primary Language**: Java
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2024-04-27
- **Last Updated**: 2024-04-27

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# VideoSynopsis
Web-based video synopsis system using Apache Spark
基于Apache Spark的Web视频摘要系统

Surveillance videos are big data, and people are usually interested in the moving targets of them. Since surveillance cameras often record with fixed scenes as backgrounds, the generated videos contain long period of clips without moving targets. Finding moving targets in such video clips will be manpower and time consuming. This project focuses on implementing a system using video synopsis algorithm to recombine the moving targets in a long surveillance video into a short video clip containing all moving targets.
视频通常以固定的场景为背景进行录制，生成的视频中包含了长时间的没有移动目标的片段。在如此长的视频片段中寻找移动目标将是耗费人力和时间的。本项目旨在实现一个系统，使用视频摘要算法将一个长监控视频中的所有移动目标重新组合成一个短片段。

The project proposes a system that can be divided into two parts: the frontend web pages and the backstage processing algorithm based on Apache Spark. In the web pages, users can upload as well as download video clip. This project designs and implements a synopsis algorithm in the backstage which uses background subtraction to grab all the frames containing moving targets in a video and then concatenate all the frames into a new shorter clip. The algorithm connects to Spark cluster. After users upload a video to be condensed on web page, the algorithm will process it and return success information to users after generating the new video. Then users can download the new video from the download page.
该项目提出了一个可以分为两部分的系统：前端网页和基于Apache Spark的后台处理算法。在网页端，用户可以上传和下载视频片段。该项目在后台设计并实现了一个摘要算法，该算法使用背景减除技术从视频中提取所有包含移动目标的帧，然后将所有帧拼接成一个更短的剪辑。该算法连接到Spark集群。用户在网页端上传需要压缩的视频后，算法将在生成新视频后处理该视频并向用户返回成功信息。然后用户可以在下载页面下载新视频。