News video story unit correlation method

Abstract

The invention discloses an association method of news video story units, aiming at providing a method for associating the story units and improving the speed and the accuracy for identifying similar key frames on the basis of local key points. The technical proposal is that: firstly, preprocessing is carried out to collected news videos to be processed, a sub-database is constructed according to a time selection strategy, the scene of an announcer is removed and the scene classification of the key frames is carried out; secondly, the local key points are probed and described by utilizing a Gauss difference DOG local point probing method and an SIFT feature description method so as to obtain a local point collection; thirdly, similar key frame identification is carried out by adopting a level filtration method so as to obtain the similar key frames; finally, judgment is carried out to the association relation between the story units so as to obtain the association relation between the story units in the sub-database and between different sub-databases. By adopting the method, the accuracy and the speed of the identification can be improved, and the requirements of users on the tracking, the organization and the checking of news video data can be satisfied.

Description

technical field [0001] The invention relates to a method for associating news video story units in the technical field of multimedia information processing. Its essence is a method for identifying and associating similar key frames between story units, and it is a method that can adapt to a variety of news from different sources and in different languages. Video processing method. Background technique [0002] With the rapid development of information technology and network technology, the sources of news videos are becoming more and more extensive, the number is increasing, and the acquisition is more convenient. There are tons of news videos from different countries, languages, TV stations and news sites all the time. In order to facilitate users to access the news information they need at any time, it is necessary to process, organize and label news videos. At present, these processes are completely dependent on manual work, which is a very time-consuming task. At the s...

AI Technical Summary

Problems solved by technology

[0007] (2) Acceleration methods such as indexing are very effective for static databases, but in the face of dynamic news video databases in practical applications, repeated indexing is a complex and time-consuming task

Although the indexing and other methods used in the current research can speed up data reading, they have not essentially solved the speed problem of image matching analysis.

[0008] (3) There are a large number of similar key frames in the same program and TV announcer shots in the news video, and they are not obviously helpful for analyzing the similarity of news stories

[0011] The existing research on news video story unit association can not effectively meet the needs of assisting users in tracking, organizing and retrieving news events. It is particularly important to study a fast and effective news video story unit association method

Images