Fire detection using statistical color model in video sequences

1
Fire detection using statistical color model in
video sequences
Journal of Visual Communication and Image
Representation , Volume 18, Issue 2, 2007,
Pages 176-185Turgay Celik, Hasan Demirel,
Huseyin Ozkaramanli, Mustafa Uyguroglu

• Adviser Yu-Chiang Li
• Speaker Wei-Cheng Wu
• Date 2009/05/12

2
Outline

• Introduction
• Background modeling
• Statistical fire color model
• Combining color and background subtraction
• Performance analysis and computational complexity
• Conclusions

3
1. Introduction

• An algorithm combines color information of fire
  with temporal changes, and background subtraction
  assisted foreground object segmentation to detect
  fire.
• Simple adaptive background model of the scene is
  generated by using three Gaussian distributions,
  where each distribution corresponds to the pixel
  statistics in the respective color channel.
• The foreground information verified by the
  statistical fire color model to determine whether
  the detected foreground object is a fire
  candidate or not.

4
2. Background modeling
5
2. Background modeling

• Change map

is the global constant
6
2. Background modeling
7
2. Background modeling
8
3. Statistical fire color model
9
3. Statistical fire color model
10
3. Statistical fire color model
11
3. Statistical fire color model
12
3. Statistical fire color model
13
4. Combining color and background subtraction
14
4. Combining color and background subtraction
15
4. Combining color and background subtraction
16
4. Combining color and background subtraction
17
5. Performance analysis and computational
complexity
18
6. Conclusions

• A real-time fire-detector, which combines color
  information with registered background scene.
• The correct detection rate for our algorithm is
  98.89.
• The proposed algorithm can be extended to
  incorporate the smoke in the video sequences,
  which may be used as faster fire alarm detection
  in such special conditions.