Pyramid histograms of motion context with application to angiogram video classification

Fei Wang, Yong Zhang, David Beymer, Hayit Greenspan, Tanveer Syeda-Mahmood

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Due to poor image quality as well as the difficulty of modeling the non-rigid heart motion, motion information has rarely been used in the past for angiogram analysis. In this paper we propose a new motion feature for the purpose of classifying angiogram videos according to their viewpoints. Specifically, local motion content of the video around the anatomical structures cardiac vessels is represented using the so-called "motion context", a motion histogram representation in polar coordinates within a local patch. The global motion layout is captured as pyramid histograms of the motion context (PHMC) in a manner similar to that proposed by the Spatial Pyramid Kernel [1]. The PHMC is a robust representation of the motion features in a video sequence. Through experiments on a large database of angiograms obtained from both diseased and control subjects, we show that our technique consistently outperforms state-of-the-art methods in the angiogram classification test.

Original languageEnglish
Title of host publicationFunctional Imaging and Modeling of the Heart - 6th International Conference, FIMH 2011, Proceedings
Pages384-391
Number of pages8
DOIs
StatePublished - 2011
Event6th International Conference on Functional Imaging and Modeling of the Heart, FIMH 2011 - New York City, NY, United States
Duration: 25 May 201127 May 2011

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume6666 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference6th International Conference on Functional Imaging and Modeling of the Heart, FIMH 2011
Country/TerritoryUnited States
CityNew York City, NY
Period25/05/1127/05/11

Fingerprint

Dive into the research topics of 'Pyramid histograms of motion context with application to angiogram video classification'. Together they form a unique fingerprint.

Cite this