Video Deduplication Based on Perceptual Hash
雪晴 胡 · Software Engineering and Applications · 2018
随着人类信息技术的不断发展，整个互联网中，各行各业累积的数据量越来越大，在云端的多媒体数据中存在着大量的冗余，同时又不断的通过海量客户端，上传新的重复数据，浪费了大量的带宽。因此如何高效去重成为亟待解决的问题。本文主要研究多媒体数据去重领域中的视频去重。提出了一种基于感知哈希的自适应阈值关键帧提取方法。并在此基础上提出了一个视频去重的方案，该方案包括关键帧提取，匹配排序和视频质量比较。并通过实验证实该方案具有较高的准确率。 With the continuous development of human information technology, the total amount of data accumulated in all walks of life is becoming larger and larger in the whole Internet. There is a lot of redundancy in the multimedia data at the cloud end. At the same time, a lot of new duplicates are uploaded through massive clients, and a lot of bandwidth is wasted. Therefore, how to effectively re-emphasize it has become an urgent problem. This paper focuses on video deduplication in multimedia data deduplication. An adaptive threshold key frame extraction method based on perceptual Hashi is proposed. Based on this, a video de duplication scheme is proposed, which includes key frame extraction, matching sorting and video quality comparison. Experiments confirm that the scheme has high accuracy.