Method for retrieving similar video clips based on sports competition videos

Abstract

The invention relates to a method for retrieving similar video clips based on sports competition videos. The method comprises the following steps of: preprocessing all videos in a video library by using an off-line part to form a video feature library; performing video structured analysis and feature extraction on the queried clips by using an on-line part; and performing two-round vari-size grained retrieval in the video feature library, wherein a candidate video set is determined in the first-round coarse-grained retrieval process, and the accurate positions of the similar clips in a video object are determined in the second-round fine-grained retrieval process. The query process is accelerated by organizing all video features through a K-dimension (K-D) tree; and the influence of vision, a time sequence and an interference factor on similarities is comprehensively considered, and the similarities of the similar chips are calculated and sequenced.

Description

technical field [0001] The invention relates to a method for retrieving similar video segments, belonging to the technical field of computer vision. Background technique [0002] Similar video retrieval is a research hotspot in the field of content-based video retrieval (CBVR) in recent years. Because of its wide application and the ability to provide key technologies for video copy detection and video semantic annotation, it has become the focus of many research institutions and scholars. Among them, the automatic retrieval of sports game videos has high practical value. The research results can be used by professionals to quickly analyze the technical actions and tactics of players in sports games, thereby improving the efficiency and level of sports training. It can also be used by sports enthusiasts to more accurately, conveniently and quickly retrieve interesting video clips, thereby improving the convenience for people to watch sports games. Similar video retrieval i...

AI Technical Summary

Problems solved by technology

[0003] At present, in practical applications, it is based on single-round retrieval, and directly calculates the similarity of all videos in the video library, but there is a serious problem in direct retrieval: since video contains text, sound and images, it is a complex Unstructured streaming data has high feature dimension, large amount of data, and complex calculation. If the video library is directly processed, the retrieval time will inevitably be too long, and it is difficult to meet the requirements of real-time query.

In addition, in view of the high dimensionality of video features and the large amount of data and complex calculations, the current similar segment retrieval methods often use some common dimensionality reduction methods (such as VA-File) to avoid the "dimension disaster" for the organization of video features. this serious problem

Commonly used dimensionality reduction methods can reduce time complexity and calculation time to a certain extent, but dimensionality reduction methods are based on the principle of approximate vectors, which will lead to reduced retrieval accuracy, and some commonly used dimensionality reduction methods have their scope of application, while video Features are random and difficult to meet the requirements

Images