基于视频七阶距特征的内容检索方法  

Video Contents Retrieval Based on Seven-order Moments

作　　者：李炎[1] 杜秀华[1] 曹俊 

机构地区：[1]上海交通大学自动化系,系统控制与信息处理教育部重点实验室,上海200240 [2]上海东方娱乐传媒集团广告经营中心,上海200041

出　　处：《实验室研究与探索》2012年第12期39-42,127,共5页Research and Exploration In Laboratory

摘　　要：为了能准确地在视频媒体中寻找到所需要的信息片段,提出了一种基于视频图像七阶矩特征的内容检索方法。通过计算图像的七阶矩特征,并以此替代视频每帧图像的信息,对视频中每帧图像及其前一帧的七阶矩特征向量进行比对,计算两者距离,当两者欧氏距离超过某一给定阈值时记录下该帧在视频中的位置,且定义该帧为视频的一个关键帧,并以此为据建立视频文件的带时间标记的关键帧七阶矩向量序列。继而通过对视频流和目标视频的七阶矩序列位置的比对和对应七阶矩距离的计算,确定在视频流中是否包含目标视频,以此实现在视频流中对于目标视频的检索。使用实际视频为例对方法进行验证,效果良好,结果准确,方法还具有较快处理速度和较少的内存消耗,可用于实时处理领域。To accurately find the needed information in the video,an approach based on the seven-order moments of video frame image was introduced in this paper.By computing the seven-order moments vector as the characteristic of an image,and replacing every frame of image with those moments vectors,every frame image of a video was compared with the previous frame,and the distance between the two was computed.The moments vector with the place was marked in the video when their distance exceeds a certain value,and such a frame was defined as a key frame of this video.Then a seven-order moments vector sequence with time marks that describes the key frames of the video was derived.By comparing the time marks and computing the distance between the corresponding moments vectors of the vector sequences of the video stream and the target video,it can be determined if the target video is in the video stream,and the target video can be detected successfully in the video stream.This approach proves to have good performance and accuracy after being tested with real examples.Meanwhile,it has very high processing speed and very little memory consumption.

关 键 词：视频处理 镜头检测 关键帧 七阶矩 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]