Fire detection method and apparatus

• US 7,680,297 B2
• Filed: 05/17/2005
• Issued: 03/16/2010
• Est. Priority Date: 05/18/2004
• Status: Active Grant

• Current Assignee: Axonx Fike Corporation
• Original Assignee: Axonx Fike Corporation
• Inventors: Privalov, George
• Primary Examiner(s): Mehta; Bhavesh M
• Assistant Examiner(s): Akhavannik; Hadi
• Application Number:
  US11/130,880
• Publication Number:
  US 20050271247A1
• Time in Patent Office: 1,764 Days
• Field of Search: 382/100, 340/578
• US Class Current: 382/100
• CPC Class Codes: G06V 20/40 in video content extracting overlay text G06V20/62; video retrieval G06F16/70; processing of video elementary streams in video servers H04N21/234; processing of video elementary streams in video clients H04N21/44

Abstract

The present invention provides a method and apparatus for detecting fire in a monitored area even if the flames are hidden behind obstructing objects. The steps of this method include: (a) detecting and capturing, at a prescribed frequency, video images of the monitored area, (b) converting the captured images into two-dimensional bitmaps of the temporally varying brightness values observed in the captured images, wherein the spatial resolution of this bitmap is determined by the number of pixels comprising the bitmaps, (c) specifying for any two of the pixels in the bitmaps a temporal coherence factor whose magnitude is a measure over a prescribed time of the similarities observed in the temporal variations of the brightness values being captured at each of the bitmap'"'"'s pixels, (d) computing this temporal coherence factors for each of the prescribed set of pixels in the captured images, and (e) analyzing the computed temporal coherence factors to identify those sets of pixels that have temporal coherence factors whose values are above a prescribed threshold value, wherein this prescribed threshold value is set so as to identify those pixels that correspond to regions of the monitored area whose temporal variations in brightness indicate that the source of this brightness is a fire in the monitored area.

Claims

1. A method of fire detection in a monitored area comprising the steps of:

detecting and capturing, at a prescribed frequency, video images of said area using a video imaging and capture device,

converting said captured images into two-dimensional brightness value bitmaps of temporally varying brightness values observed in said images using a computer system, wherein spatial resolution of said bitmaps being determined by a number of pixels comprising said bitmaps,

converting said brightness value bitmaps into a two-dimensional matrix of temporal similarities between variations of the brightness values captured at each of said pixels and variations of the brightness values captured at a pixel that is adjacent to said pixel for each pixel in said brightness value bitmaps and over a prescribed time period using said computer system, and

analyzing said matrix to identify clusters of pixels that have temporal similarities whose values are above a prescribed threshold value using said computer system,

sending a signal to an alarm means to trigger said alarm means when clusters having temporal similarities whose values are above the prescribed threshold value are identified by said computer system,

wherein said prescribed threshold value is set so as to identify those clusters of pixels that correspond to regions of said monitored area whose temporal variations in adjoining pixel brightness values indicate that the source of said temporal brightness variations is a fire in said monitored area.

2. A method of fire detection as recited in claim 1, wherein said conversion of said brightness value bitmaps to said matrix involves computing for any two pixels in said brightness value bitmaps a convolution of a time derivative of the brightness values of said pixels.

3. A method of fire detection as recited in claim 1, wherein said conversion of said brightness value bitmaps to said matrix involves computing for any pair of pixels in said brightness value bitmaps a sum across a prescribed number of samples of products of brightness differences of said pixels in successive bitmaps.

4. A method of fire detection as recited in claim 1, wherein said conversion of said brightness value bitmaps to said matrix involves for any pair of pixels in said brightness value bitmaps a timed-averaged sum of products of brightness differences of said pixels in successive bitmaps.

5. A method of fire detection as recited in claim 1, wherein said conversion of said brightness value bitmaps to said matrix involves computing for any pair of pixels in said brightness value bitmaps a sum across a prescribed number of samples of products of differences in current brightness value and a time-averaged brightness value of said pixels.

6. A method of fire detection as recited in claim 1, wherein said conversion of said brightness value bitmaps to said matrix involves computing for any pair of pixels in said brightness value bitmaps the time-averaged sum of products of differences in current brightness value and the time-averaged brightness value of the said pixels.

7. A method of fire detection as recited in claim 1, wherein said conversion of said brightness value bitmaps to said matrix involves computing for any pair of pixels in said brightness value bitmaps a sum of products of Fourier terms of a time derivative brightness of one said pixel and complex conjugate Fourier terms of another pixel.

8. A method of fire detection as recited in claim 1, wherein said conversion of said brightness value bitmaps to said matrix involves computing for any pair of pixels in said brightness value bitmaps a sum of a product of non-zero Fourier terms of a brightness value of one said pixel and complex conjugate Fourier terms of a brightness value of another pixel performed over a prescribed range of terms.

9. A method of fire detection as recited in claim 1, wherein said clusters having a measurable size, and further comprising the step of triggering said alarm means when one of said clusters has exceed a prescribed size.

10. An apparatus for fire detection within a monitored area, said apparatus comprising:

a means for detecting and capturing, at a prescribed frequency, video images of said area,

a means for converting said captured images into two-dimensional brightness value bitmaps of temporally varying brightness values observed in said images, wherein a spatial resolution of said bitmaps being determined by a number of pixels comprising said bitmaps,

a means for converting said brightness value bitmaps into a two-dimensional matrix of temporal similarities between variations of the brightness values captured at each of said pixels and variations of the brightness values captured at a pixel that is adjacent to said pixel for each pixel in said brightness value bitmap and over a prescribed time period, and

a means for analyzing said matrix to identify clusters of pixels that have temporal similarities whose values are above a prescribed threshold value,

wherein said prescribed threshold value is set so as to identify those clusters of pixels that correspond to regions of said monitored area whose temporal variations in adjoining pixel brightness values indicate that the source of said temporal brightness variations is a fire in said monitored area.

11. An apparatus for fire detection as recited in claim 10, wherein said conversion of said brightness value bitmaps to said matrix involves computing for any two pixels in said brightness value bitmaps a convolution of a time derivative of the brightness values of said pixels.

12. An apparatus for fire detection as recited in claim 10, wherein said conversion of said brightness value bitmaps to said matrix involves computing for any pair of pixels in said brightness value bitmaps a sum across a prescribed number of samples of products of brightness differences of said pixels in successive bitmaps.

13. An apparatus for fire detection as recited in claim 10, wherein said conversion of said brightness value bitmaps to said matrix involves for any pair of pixels in said brightness value bitmaps a timed-averaged sum of products of brightness differences of said pixels in successive bitmaps.

14. An apparatus for fire detection as recited in claim 10, wherein said conversion of said brightness value bitmaps to said matrix involves computing for any pair of pixels in said brightness value bitmaps a sum across a prescribed number of samples of products of differences in current brightness value and a time-averaged brightness value of said pixels.

15. An apparatus for fire detection as recited in claim 10, wherein said conversion of said brightness value bitmaps to said matrix involves computing for any pair of pixels in said brightness value bitmaps the time-averaged sum of products of differences in current brightness value and the time-averaged brightness value of the said pixels.

16. An apparatus for fire detection as recited in claim 10, wherein said conversion of said brightness value bitmaps to said matrix involves computing for any pair of pixels in said brightness value bitmaps a sum of products of Fourier terms of a time derivative brightness of one said pixel and complex conjugate Fourier terms of another pixel.

17. An apparatus for fire detection as recited in claim 10, wherein said conversion of said brightness value bitmaps to said matrix involves computing for any pair of pixels in said brightness value bitmaps a sum of a product of non-zero Fourier terms of a brightness value of one said pixel and complex conjugate Fourier terms of a brightness value of another pixel performed over a prescribed range of terms.

18. An apparatus for fire detection as recited in claim 10, wherein said clusters having a measurable size, and further comprising a means for issuing an alarm that signifies when one of said clusters has exceed a prescribed size.