Weather-adaptive flying target detection and tracking from infrared video sequences

Tze Lun Weng, Yi Ying Wang, Zong Ying Ho, Yung Nien Sun

Research output: Contribution to journalArticlepeer-review

12 Citations (Scopus)

Abstract

This paper proposes a new computer-aided tracking method that can track multiple flying targets in an image sequence acquired from either an optical camera or an infrared camera system. The proposed method includes two major steps, which are flying target detection and flying target tracking. Because weather conditions greatly affect the detection of a flying target, we designed a fuzzy system that classifies the weather into two basic weather types. Based on the weather type, flying targets can be detected successfully. In tracking a flying target, we apply the dynamic layer representation method to represent the flying target and use a Kalman filter to predict the target's position. The second fuzzy rule system is designed to identify the target states, which are visible, missing, or occluded, based on the layer and positional features. Thereafter, the identified state can help to select a suitable tracking strategy for the subsequent frames. In the experiments, we demonstrate the results applied to five real image sequences. The proposed method can successfully detect and track multiple flying targets perfectly in clear weather. It also achieves around 90% accuracy when the weather is not clear. The fuzzy methodology makes the proposed system more effective at detecting and tracking targets and more flexible to complex environmental conditions, since the rules can be easily added and modified.

Original languageEnglish
Pages (from-to)1666-1675
Number of pages10
JournalExpert Systems With Applications
Volume37
Issue number2
DOIs
Publication statusPublished - 2010 Mar 1

All Science Journal Classification (ASJC) codes

  • Engineering(all)
  • Computer Science Applications
  • Artificial Intelligence

Fingerprint Dive into the research topics of 'Weather-adaptive flying target detection and tracking from infrared video sequences'. Together they form a unique fingerprint.

Cite this