Tennis video analysis based on transformed motion vectors

Peng Wang, Rui Cai, Shi Qiang Yang

Research output: Chapter in Book/Report/Conference proceedingChapterpeer-review

Abstract

Motion Vectors (MV) indicate the motion characteristics between two video frames, and has been widely used in the contentbased sports video analysis. Previous works on sports video analysis have proved the effectiveness and efficiency of the MV-based methods. However, in the tennis video, the MV-based methods are seldom applied because the motion represented by MV is greatly deformed relative to the player's true movement due to the camera's diagonal shooting. In this paper, an algorithm of MV transformation is proposed to revise the deformed MV using a pinhole camera model. With the transformed MVs, we generate the temporal feature curves and employ Hidden Markov Models to classify two types of player's basic actions. Evaluation on four hours live tennis videos shows very encouraging results.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
EditorsPeter Enser, Yiannis Kompatsiaris, Noel E. O’Connor, Alan F. Smeaton, Arnold W. M. Smeulders
PublisherSpringer Verlag
Pages79-87
Number of pages9
ISBN (Print)3540225390, 9783540225393
DOIs
StatePublished - 2004
Externally publishedYes

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume3115
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Fingerprint

Dive into the research topics of 'Tennis video analysis based on transformed motion vectors'. Together they form a unique fingerprint.

Cite this