Structure tensor series-based matching for near-duplicate video retrieval
Structure tensor series-based matching for near-duplicate video retrieval
复制标题
DOI：
10.1145/2072298.2071937
复制
发表时间：
2011-11
期刊：
影响因子：
--
通讯作者：
Xiangmin Zhou;Lei Chen;Xiaofang Zhou
中科院分区：
文献类型：
--
作者：
Xiangmin Zhou;Lei Chen;Xiaofang Zhou
近重复视频检索因其广泛的应用而受到广泛关注，包括版权检测、商业监控和新闻视频跟踪等。近年来，有显着的研究工作，有效地识别附近的重复从大型视频集合。然而，现有的方法对大型视频数据库遭受低的准确性，由于严重的信息丢失。本文提出了一种基于三维结构张量模型的解决方案。首先，我们提出了一种新的视频表示方案，自适应结构视频张量系列（ASVT系列），连同一个强大的相似性度量，以提高检索效率。通过对数百小时真实的视频数据的大量实验，证明了该方法的有效性。
Near duplicate video retrieval has attracted much attention due to its wide spectrum of applications including copyright detection, commercial monitoring and news video tracking. In recent years, there has been significant research effort on efficiently identifying near duplicates from large video collections. However, existing approaches for large video databases suffer from low accuracy due to the serious information loss. In this paper, we propose a practical solution based on 3D structure tensor model for this problem. We first propose a novel video representation scheme, adaptive structure video tensor series (ASVT series), together with a robust similarity measure, to improve the retrieval effectiveness. Then, we prove the effectiveness of the proposed method by extensive experiments on hundreds hours real video data.