System and method for identifying location of flame within field of view

Abstract

A flame detection system is disclosed. The flame detection system includes one or more processors coupled to at least one flame detector and at least one image capture device. The one or more processors identify a location of the flame by determining a base Y-coordinate, a tip Y-coordinate, a leftmost X-coordinate, and a rightmost X-coordinate of a plurality of pixels associated with the flame; determining that the variation in the base Y coordinate of the most current image in the sequence of images is less than a predefined limit of the height of the flame; and determining that the change in the tip Y coordinate of the most current image in the sequence of images is greater than a predefined limit of the height of the flame, or the change in the leftmost X coordinate or the rightmost X coordinate of the most current image in the sequence of images is greater than a predefined limit of the width of the flame.

Description

Technical Field

[0001] The exemplary embodiments disclosed herein relate generally to flame detection systems, and more specifically to systems and methods for identifying the location of a flame within a field of view (FoV). Background Technology

[0002] In flame detection systems, flame detectors are designed to identify the presence of a flame using various technologies, such as infrared, ultraviolet, or visible light sensors. Existing flame detectors often cannot pinpoint the exact location of the flame source (e.g., indicate the location on a screen), making it difficult to identify the location of the fire when an alarm is triggered.

[0003] The inventors have identified numerous areas for improvement in the prior art and methods, which are the subject of the embodiments described herein. Through effort, ingenuity, and innovation, many of these deficiencies, challenges, and problems have been addressed by developing solutions included in the embodiments of this disclosure, some examples of which are described in detail herein. Summary of the Invention

[0004] The following is a simplified overview to provide a basic understanding of some aspects of this disclosure. This invention is not an exhaustive summary and is neither intended to identify key or essential elements nor to describe the scope of such elements. Its purpose is to serve as a prelude to the detailed embodiments provided below, to present some concepts of the described features in a simplified form.

[0005] In an example embodiment, a flame detection system is disclosed. The flame detection system includes at least one flame detector configured to generate data corresponding to the detected presence of a flame within a field of view (FoV). The flame detection system also includes at least one image capture device operatively coupled to the at least one flame detector. The at least one image capture device is configured to capture a sequence of images of the flame within the FoV. Multiple pixels in each image of the image sequence are associated with the flame within the FoV. Furthermore, the flame detection system includes one or more processors communicatively coupled to the at least one flame detector and the at least one image capture device. The one or more processors are configured to identify the location of the flame within the FoV by determining, for each image in the image sequence, the base Y-coordinate, tip Y-coordinate, leftmost X-coordinate, and rightmost X-coordinate of the multiple pixels associated with the flame within the FoV. The one or more processors are configured to identify the position of the flame within the FoV by determining that the change in the base Y-coordinate of the most current image in the image sequence compared to a previous image in the image sequence is less than a predefined limit for the height of the flame. Furthermore, the one or more processors are configured to identify the position of the flame within the FoV by determining that the change in the tip Y-coordinate of the most current image in the image sequence compared to a previous image in the image sequence is greater than the predefined limit for the height of the flame, or that the change in the leftmost or rightmost X-coordinate of the most current image in the image sequence compared to a previous image in the image sequence is greater than a predefined limit for the width of the flame.

[0006] In some embodiments, the predefined limit of the height of the flame corresponds to 2% of the height of the flame, and the predefined limit of the width of the flame corresponds to 2% of the width of the flame.

[0007] In some embodiments, the one or more processors are further configured to store the image sequence in a rolling buffer in memory at a predefined number of frames per second (FPS). Furthermore, the one or more processors are configured to determine whether the flame exists within an image in the image sequence stored in the rolling buffer. Additionally, the one or more processors are configured to, upon determining that the flame exists within an image in the image sequence stored in the rolling buffer, save that image as a previous frame. The one or more processors are further configured to store another image sequence received from the at least one image capture device in the rolling buffer in memory as a current frame at the predefined FPS.

[0008] In some implementations, the predefined FPS defines a range between 30 FPS and 60 FPS.

[0009] In some implementations, the one or more processors are configured to determine the base Y-coordinate, the tip Y-coordinate, the leftmost X-coordinate, and the rightmost X-coordinate by extracting red-green-blue (RGB) frames from the previous image and the current image in the image sequence. Furthermore, the one or more processors are configured to determine the base Y-coordinate, the tip Y-coordinate, the leftmost X-coordinate, and the rightmost X-coordinate by dividing the RGB frames of the previous image and the current image into R channels, G channels, and B channels. Furthermore, the one or more processors are configured to determine the base Y-coordinate, the tip Y-coordinate, the leftmost X-coordinate, and the rightmost X-coordinate by subtracting one or more pixels from the G channel from one or more pixels in the R channel. Furthermore, the one or more processors are configured to determine the base Y-coordinate, the tip Y-coordinate, the leftmost X-coordinate, and the rightmost X-coordinate by determining that the difference between the pixel values ​​of the R channel and the G channel is greater than 60. The one or more processors are configured to determine the base Y coordinate, the tip Y coordinate, the leftmost X coordinate, and the rightmost X coordinate by adding the one or more pixels of the R channel within the boundary of the flame to dilate the image sequence from the previous frame and the current frame. Furthermore, the one or more processors are configured to determine the base Y coordinate, the tip Y coordinate, the leftmost X coordinate, and the rightmost X coordinate by performing contour extraction on the dilated image sequence by concatenating the one or more pixels of the R channel.

[0010] In some implementations, the plurality of pixels in the image sequence correspond to one or more pixels in the current frame and one or more pixels in the previous frame.

[0011] In some implementations, this data corresponds to infrared (IR) sensor data captured at a rate of 60 samples per second.

[0012] In some embodiments, the flame detection system is further configured to indicate the location of the flame within the FoV on at least one image in the image sequence.

[0013] In another example embodiment, a flame detection system is disclosed. The flame detection system includes at least one flame detector configured to generate data corresponding to the detection of a flame within a field of view (FoV). The flame detection system also includes at least one image capture device operatively coupled to the at least one flame detector, wherein the at least one image capture device is configured to capture a sequence of images of the flame within the FoV. The plurality of pixels in each image of the image sequence are associated with the flame within the FoV. The flame detection system also includes one or more processors communicatively coupled to the at least one flame detector and the at least one image capture device. The one or more processors are configured to identify the location of the flame within the FoV by determining, for at least one image in the image sequence, a center pixel among the plurality of pixels associated with the flame within the FoV. The one or more processors are further configured to identify the location of the flame within the FoV by determining the intensity of the center pixel and the intensities of adjacent pixels. The adjacent pixels include the pixel to the left of the center pixel, the pixel to the right of the center pixel, the pixel directly below the center pixel, and the pixel directly above the center pixel. Furthermore, the one or more processors are configured to identify the position of the flame within the FoV by determining that the intensity value of the center pixel is equal to the intensity of each of the adjacent pixels.

[0014] In some embodiments, the one or more processors are further configured to determine, for the at least one image in the image sequence, the coordinates of the center pixel among the plurality of pixels associated with the flame within the FoV by: determining a vertical line between the Y-coordinate of the flame's tip and the Y-coordinate of its base. Furthermore, the one or more processors are configured to determine, for the at least one image in the image sequence, the coordinates of the center pixel among the plurality of pixels associated with the flame within the FoV by: determining a horizontal line between the leftmost and rightmost X-coordinates of the flame. Furthermore, the one or more processors are configured to determine, for the at least one image in the image sequence, the coordinates of the center pixel among the plurality of pixels associated with the flame within the FoV by: determining the intersection of the vertical line and the horizontal line. The one or more processors may be further configured to determine, for the at least one image in the image sequence, the coordinates of the center pixel among the plurality of pixels associated with the flame within the FoV by: associating the intersection as the center pixel.

[0015] In yet another example embodiment, a flame detection system is disclosed. The flame detection system includes at least one flame detector configured to generate data corresponding to the detection of a flame within a field of view (FoV). The flame detection system also includes at least one image capture device operatively coupled to the at least one flame detector, wherein the at least one image capture device is configured to capture a sequence of images of the flame within the FoV. The plurality of pixels in each image of the image sequence are associated with the flame within the FoV. The flame detection system also includes one or more processors communicatively coupled to the at least one flame detector and the at least one image capture device. The one or more processors are configured to identify the location of the flame within the FoV by determining that the flicker frequency of the at least one flame detector is equal to the flicker frequency of the image sequence captured by the at least one image capture device.

[0016] In some embodiments, the one or more processors are further configured to determine the flicker frequency of the at least one flame detector and the flicker frequency of the image sequence by determining a single average value of one or more pixels within the contour-extracted image sequence at the predefined FPS. Furthermore, the one or more processors are further configured to determine the flicker frequency of the at least one flame detector and the flicker frequency of the image sequence by converting the single average value into a frequency domain representation.

[0017] The above description of the invention is provided merely to outline some exemplary embodiments to provide a basic understanding of some aspects of the invention. Therefore, it should be understood that the above embodiments are merely illustrative and should not be construed as limiting the scope or nature of the invention in any way. It should be understood that, in addition to those described herein, the scope of the invention covers many possible embodiments, some of which will be further described below. Attached Figure Description

[0018] Therefore, some exemplary embodiments of this disclosure have been described in general terms, and reference will be made below to the accompanying drawings, which are not necessarily drawn to scale, and in which:

[0019] Figure 1 A block diagram illustrating a flame detection system for identifying the position of a flame within a field of view (FoV) according to an example embodiment of the present disclosure is shown;

[0020] Figure 2A An example embodiment of the present disclosure illustrates the integration of at least one flame detector and at least one image capture device in a flame detection system;

[0021] Figure 2BSuspicious regions in an image sequence within the FoV according to an example embodiment of this disclosure are illustrated;

[0022] Figure 3A An image sequence in a rolling buffer is illustrated according to an example embodiment of the present disclosure;

[0023] Figure 3B An example embodiment of adding another image from an image sequence to a rolling buffer is illustrated according to an example embodiment of the present disclosure;

[0024] Figure 4 A flowchart illustrating a method for storing fire images in a memory according to an example embodiment of the present disclosure is shown;

[0025] Figure 5A The R channel, G channel, and B channel of each image in an image sequence according to a first embodiment of the present disclosure are illustrated;

[0026] Figure 5B An example is illustrated by subtracting one or more pixels from the G channel from one or more pixels in the R channel according to a first embodiment of the present disclosure;

[0027] Figure 5C An example of dilation of an image sequence according to a first embodiment of the present disclosure is illustrated;

[0028] Figure 5D An example of contour extraction of an image sequence according to a first embodiment of the present disclosure is illustrated;

[0029] Figure 6 A flame is illustrated according to an example embodiment of this disclosure;

[0030] Figure 7A The illustration shows the base Y coordinate, tip Y coordinate, leftmost X coordinate, and rightmost X coordinate of a plurality of pixels associated with the flame within the FoV for the most current image and the previous image in an image sequence according to a first embodiment of the present disclosure.

[0031] Figure 7B A table illustrating predefined limits for determining the height and width of a flame according to a first embodiment of this disclosure is provided.

[0032] Figure 8A A flowchart illustrating a method for contour extraction of an image sequence according to a first embodiment of the present disclosure is shown;

[0033] Figure 8B A flowchart illustrating a method for storing an image sequence in memory as a previous frame according to a first embodiment of the present disclosure is shown;

[0034] Figure 8C A flowchart illustrating a method for contour extraction of another image sequence according to a first embodiment of the present disclosure is shown;

[0035] Figure 8D A flowchart illustrating a method for storing an image sequence in memory as a current frame according to a first embodiment of the present disclosure is shown;

[0036] Figure 8E A flowchart illustrating a method for identifying the location of a flame according to a first embodiment of the present disclosure is shown;

[0037] Figure 9A An example is illustrated of a flame having a boundary around it according to a second embodiment of this disclosure;

[0038] Figure 9B An example is shown of a vertical line between the Y-coordinate of the tip of a flame and the Y-coordinate of its base, and a horizontal line between the leftmost X-coordinate and the rightmost X-coordinate of a flame, according to a second embodiment of this disclosure.

[0039] Figure 9C A plurality of pixels associated with a flame are illustrated according to a second embodiment of the present disclosure;

[0040] Figure 10A An image sequence is illustrated, showing that the intensity value of the center pixel according to a second embodiment of the present disclosure is equal to the intensity of each of the neighboring pixels.

[0041] Figure 10B A table illustrating an image sequence according to a second embodiment of the present disclosure shows that the intensity value of the center pixel is equal to the intensity of each of the neighboring pixels.

[0042] Figure 11A A flowchart illustrating a method for contour extraction of an image sequence according to a second embodiment of the present disclosure is shown;

[0043] Figure 11B A flowchart illustrating a method for determining the center pixel among a plurality of pixels associated with a flame within a FoV for at least one image in an image sequence, according to a second embodiment of the present disclosure;

[0044] Figure 11C A flowchart illustrating a method for identifying the location of a flame according to a second embodiment of the present disclosure is shown;

[0045] Figure 12 An example is illustrated by aligning the flashing frequency of at least one flame detector with the flashing frequency of an image sequence captured by at least one image capturing device according to a third embodiment of the present disclosure;

[0046] Figure 13 Multiple graphs illustrating the flicker frequency of at least one flame detector according to a third embodiment of the present disclosure and the flicker frequency of an image sequence captured by at least one image capturing device are shown.

[0047] Figure 14A A flowchart illustrating a method for contour extraction of an image sequence according to a third embodiment of the present disclosure is shown;

[0048] Figure 14B A flowchart illustrating the steps for determining the flash frequency of at least one flame detector and the flash frequency of an image sequence according to a third embodiment of the present invention is shown; and

[0049] Figure 14C A flowchart illustrating a method for identifying the location of a flame according to a third embodiment of the present disclosure is shown. Detailed Implementation

[0050] Some embodiments will now be described more fully below with reference to the accompanying drawings, which illustrate some embodiments, but not all embodiments. In fact, various embodiments may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided to enable this disclosure to meet applicable legal requirements.

[0051] The components illustrated in the accompanying drawings represent components that may or may not be present in the various embodiments of the invention described herein, such that embodiments may include fewer or more components than those shown in the figures without departing from the scope of the invention. Some components may be omitted from one or more figures, or shown in dashed lines to make components visible below.

[0052] As used herein, the term “comprising” means including but not limited to, and should be interpreted in the manner in which it is typically used in the patent context. The use of broader terms such as “comprising,” “including,” and “having” should be understood to provide support for narrower terms such as “consisting of,” “substantially composed of,” and “substantially constituted by.”

[0053] The phrases “in various embodiments,” “in one embodiment,” “according to one embodiment,” “in some embodiments,” etc., generally mean that the specific feature, structure, or characteristic following the phrase may be included in at least one embodiment of this disclosure, and may be included in more than one embodiment of this disclosure (importantly, such phrases do not necessarily refer to the same embodiment).

[0054] As used herein, the terms “example” or “exemplary” mean “serving as an example, instance, or illustration.” Any specific implementation described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other specific implementations.

[0055] If this specification states that a component or feature is "may", "can", "may", "should", "will", "preferably", "possibly", "typically", "optionally", "for example", "usually", or "may" (or other such language) included or has a characteristic, then the specific component or feature does not need to be included or have that characteristic. Such components or features may be optionally included in some embodiments, or they may be excluded.

[0056] This disclosure provides various embodiments of systems and methods for identifying the location of a flame within a field of view (FoV). Each embodiment can be configured to generate data corresponding to the detected flame within the FoV. Furthermore, each embodiment can be configured to capture an image sequence of the flame within the FoV. Each embodiment can be configured to, for each image in the image sequence, determine the base Y-coordinate, tip Y-coordinate, leftmost X-coordinate, and rightmost X-coordinate of a plurality of pixels associated with the flame within the FoV. Each embodiment can be configured to determine that the change in the base Y-coordinate of the most current image in the image sequence compared to previous images in the image sequence is less than a predefined limit for the height of the flame. Each embodiment can be further configured to determine that the change in the tip Y-coordinate of the most current image in the image sequence compared to previous images in the image sequence is greater than the predefined limit for the height of the flame, or that the change in the leftmost or rightmost X-coordinate of the most current image in the image sequence compared to previous images in the image sequence is greater than a predefined limit for the width of the flame.

[0057] Each embodiment can be configured to: store the image sequence in a rolling buffer in memory at a predefined number of frames per second (FPS). Each embodiment can be configured to: determine whether the flame is likely to exist within an image in the image sequence stored in the rolling buffer. Furthermore, each embodiment can be configured to: when it is determined that the flame is likely to exist within an image in the image sequence stored in the rolling buffer, save that image as a previous frame. Additionally, each embodiment can be configured to: store another image sequence received from the at least one image capture device in the rolling buffer in memory as a current frame at the predefined FPS.

[0058] Each embodiment can be configured to: determine the base Y-coordinate, the tip Y-coordinate, the leftmost X-coordinate, and the rightmost X-coordinate. Each embodiment can be further configured to: extract red-green-blue (RGB) frames from the previous image and the current image in the image sequence. Each embodiment can be further configured to: divide the RGB frames of the previous image and the current image into R channels, G channels, and B channels. Furthermore, each embodiment can be further configured to: subtract one or more pixels of the G channel from one or more pixels of the R channel. Each embodiment can be configured to: determine that the difference in pixel values ​​between the R channel and the G channel is greater than 60. Each embodiment can be configured to: dilate the image sequence from the previous frame and the current frame by adding one or more pixels of the R channel within the boundary of the flame. Furthermore, each embodiment can be further configured to: extract the contour of the dilated image sequence by concatenating one or more pixels of the R channel. Each embodiment can be further configured to: indicate the position of the flame within the FoV on at least one image in the image sequence.

[0059] Each embodiment can be configured to: for at least one image in the image sequence, determine a center pixel among the plurality of pixels associated with the flame within the FoV. Each embodiment can be configured to: determine the intensity of the center pixel and the intensities of its neighboring pixels. Furthermore, each embodiment can be further configured to: determine that the intensity value of the center pixel is equal to the intensity of each of its neighboring pixels.

[0060] Each embodiment can be configured to: for the at least one image in the image sequence, determine the coordinates of the center pixel among the plurality of pixels associated with the flame within the FoV. Each embodiment can be configured to: determine a vertical line between the Y-coordinate of the flame's tip and the Y-coordinate of its base. Each embodiment can be further configured to: determine a horizontal line between the leftmost and rightmost X-coordinates of the flame. Each embodiment can be further configured to: determine the intersection of the vertical line and the horizontal line. Furthermore, each embodiment can be configured to: associate the intersection point with the center pixel.

[0061] Each embodiment can be configured to: determine that the flashing frequency of the at least one flame detector is equal to the flashing frequency of the image sequence captured by the at least one image capture device. Each embodiment can be further configured to: determine a single average value of one or more pixels within the contour-extracted image sequence at a predefined FPS. Each embodiment can be further configured to: convert the single average value into a frequency domain representation.

[0062] Figure 1A block diagram of a flame detection system 100 for identifying the position of a flame within a field of view (FoV) 116 according to an example embodiment of the present disclosure is shown. Figure 2A An example embodiment of the present disclosure is illustrated in which at least one flame detector 102 and at least one image capture device 104 are integrated in a flame detection system 100.

[0063] In some embodiments, the flame detection system 100 may include at least one flame detector 102, at least one image capture device 104, and one or more processors 106. The at least one flame detector 102 and the at least one image capture device 104 may be placed at a location. The at least one flame detector 102 may be configured to generate data corresponding to the detected presence of a flame within a field of view (FoV) 116. The at least one image capture device 104 is operatively coupled to the at least one flame detector 102. The at least one image capture device 104 may be configured to capture a sequence of images of the flame within the FoV 116. Multiple pixels of each image in the image sequence may be associated with a flame within the FoV 116.

[0064] In some embodiments, at least one flame detector 102 may be responsible for detecting the presence of a flame within the FoV 116 using multiple sensors. These multiple sensors can identify unique characteristics of the flame. These unique characteristics may correspond to infrared light emission, ultraviolet light emission, or visible light emission. At least one flame detector 102 may be equipped with multiple sensors that can detect the presence of a flame by identifying optical features associated with it. These multiple sensors may be based at least on infrared (IR) light detection technology, ultraviolet (UV) light detection technology, or visible light detection technology. In some embodiments, at least one flame detector 102 may correspond to an IR sensor-based flame detector. These multiple sensors can identify the flame under one or more conditions. At least one flame detector 102 may continuously monitor the FoV 116 and may trigger an alarm when at least one flame detector 102 may detect a flame. Once a flame may be detected, at least one image capture device 104 may capture a sequence of images of the location. At least one image capture device 104 may correspond to a camera. The image sequence of the flame within the FoV 116 may be stored in a memory and may be further processed by one or more processors 106.

[0065] In some implementations, the image capture device 104 may capture a sequence of images. In one example, the image sequence may correspond to a "fire image." The image sequence may include moments prior to alarm activation. In some implementations, the process of creating a "fire image" may begin by generating data corresponding to the detection of a flame within the FoV 116. At least one flame detector 102 may be equipped with an IR sensor. The IR sensor may be able to detect specific infrared emission characteristics of the flame. When the IR sensor detects infrared emission indicating the presence of a flame, the flame detection system 100 may trigger at least one image capture device 104 to begin capturing an image sequence of the flame within the FoV 116. At least one image capture device 104 may capture an image sequence or video frames within the FoV 116. The image sequence or video frames may be stored in memory. The image sequence may include both the warning period and the moment immediately following the detection of the flame.

[0066] In some embodiments, one or more processors 106 may be configured to identify the position of a flame within a FoV 116 by performing any of the three embodiments described below. In a first embodiment, one or more processors 106 may be configured to, for each image in the image sequence, determine the base Y coordinate, tip Y coordinate, leftmost X coordinate, and rightmost X coordinate of the plurality of pixels associated with the flame within the FoV 116. Furthermore, one or more processors 106 may be configured to determine that the change in the base Y coordinate of the most current image in the image sequence compared to previous images in the image sequence is less than a predefined limit for the height of the flame. Additionally, one or more processors 106 may be configured to determine that the change in the tip Y coordinate of the most current image in the image sequence compared to previous images in the image sequence is greater than the predefined limit for the height of the flame, or that the change in the leftmost or rightmost X coordinate of the most current image in the image sequence compared to previous images in the image sequence is greater than a predefined limit for the width of the flame.

[0067] In some embodiments, the predefined limit of the flame height may correspond to 2% of the flame height, and the predefined limit of the flame width may correspond to 2% of the flame width. In some embodiments, the base Y-coordinate may correspond to the bottommost pixel coordinate of the flame. The tip Y-coordinate may correspond to the topmost pixel coordinate of the flame. The leftmost X-coordinate may correspond to the left X-coordinate corresponding to the far leftmost pixel coordinate of the flame, and the rightmost X-coordinate may correspond to the right X-coordinate corresponding to the far rightmost pixel coordinate of the flame.

[0068] In a second embodiment, one or more processors 106 may be configured to determine, for at least one image in the image sequence, a center pixel among the plurality of pixels associated with the flame within the FoV 116. Furthermore, one or more processors 106 may be configured to determine the intensity of the center pixel and the intensities of neighboring pixels. The neighboring pixels may include pixels located to the left of the center pixel, pixels located to the right of the center pixel, pixels directly below the center pixel, and pixels directly above the center pixel. Additionally, one or more processors 106 may be configured to determine that the intensity value of the center pixel is equal to the intensity of each of the neighboring pixels.

[0069] In some implementations, the intensity values ​​of neighboring pixel coordinates of the center pixel coordinate are the same as the pixel value of the center pixel coordinate in the image sequence. In some implementations, one or more processors 106 may be configured to determine whether the pixel value of the center pixel coordinate and the neighboring pixel coordinate are substantially the same. One or more processors 106 may compare the intensity and color values ​​of the neighboring pixel coordinate and the center pixel coordinate. One or more processors 106 may further improve flame detection accuracy by analyzing the pixel value within the image sequence to confirm the presence of the flame. If the neighboring pixel coordinates exhibit similar pixel values, one or more processors 106 may suggest a consistent and continuous flame, rather than random noise or reflections.

[0070] In a third embodiment, one or more processors 106 may be configured to determine that the flashing frequency of at least one flame detector 102 is equal to the flashing frequency of an image sequence captured by at least one image capture device 104. In some embodiments, one or more processors 106 are configured to align the flashing frequencies of an IR sensor and at least one image capture device 104 to synchronize data. The synchronized data can be used to determine accurate flame source location, and the IR sensor may be integrated into at least one flame detector 102. One or more processors 106 may synchronize the flashing frequency of at least one flame detector 102 with the flashing frequency of the image capture device 104 to enhance the accuracy of flame location. In some embodiments, flames may inherently exhibit characteristic flashing patterns due to the turbulent nature of combustion. In some embodiments, the flame detection system 100 can more effectively correlate data by aligning the flashing frequency of at least one flame detector 102, which detects infrared emissions, with the image sequence capture rate of at least one image capture device 104. The synchronized data provides a comprehensive view of the flame's behavior over time, thereby allowing one or more processors 106 to accurately pinpoint the flame's location.

[0071] In some embodiments, the flame detection system 100 is further configured to indicate the position of the flame within the FoV 116 on at least one image of the image sequence. Furthermore, multiple pixels in the image sequence may correspond to one or more pixels in the current frame and one or more pixels in a previous frame.

[0072] One or more processors 106 may include suitable logic, circuitry, and / or interfaces operable to execute one or more instructions stored in memory to perform predetermined operations. In one embodiment, one or more processors 106 may be configured to decode and execute any instructions received from one or more other electronic devices or servers. One or more processors 106 may be configured to execute one or more computer-readable program instructions, such as program instructions for performing any of the functions described in this specification. Furthermore, one or more processor technologies known in the art may be used to implement the processor. Examples of one or more processors 106 may include, but are not limited to, one or more general-purpose processors (e.g., or Advanced Micro (AMD) microprocessors) and / or one or more dedicated processors (e.g., digital signal processors or System-on-a-Chip (SOC) Field Programmable Gate Array (FPGA) processor.

[0073] In some embodiments, after a flame is detected using any of the three embodiments, one or more processors 106 may transmit control signals to relay 108, analog output 110, and communication 112. Relay 108, analog output 110, and communication 112 may be used as outputs of the flame detection system 100. In some embodiments, when generating data corresponding to the presence of a flame in the detected FoV 116, the flame and gas unit 114 may activate safety measures (such as alarms or shutdown procedures) to prevent danger. Furthermore, if the flame detection system 100 may include a gas detection component, the flame and gas unit 114 may monitor the presence of combustible or toxic gases.

[0074] It will be apparent to those skilled in the art that the aforementioned components of the flame detection system 100 are provided for illustrative purposes only without departing from the scope of this disclosure.

[0075] Figure 2B A suspicious region 204 in an image sequence 202 within a FoV according to an example embodiment of the present disclosure is illustrated.

[0076] In some embodiments, one or more processors 106 may mark and highlight suspicious regions 204 on the captured image sequence 202. Suspicious region 204 is a specific area within the captured image sequence 202 or video frame that is suspected of containing flames or a potential false alarm. One or more processors 106 may analyze the captured image sequence 202 to identify multiple pixels associated with flames within the FoV 116 and determine the coordinates of the identified multiple pixels. In some embodiments, once the multiple pixels are identified, one or more processors 106 may mark the suspicious region 204 on the analyzed image sequence 202. Marking of the suspicious region 204 is accomplished by drawing a rectangle or another shape around the area containing the multiple pixels. Suspicious region 204 may indicate a region of interest on the captured image sequence 202, video frame sequence, or fire imagery.

[0077] In some implementations, the suspected area 204 may correspond to an image from image sequence 202 where the location of the flame is highlighted. The markings may include visual indicators that clearly indicate the location of the flame. The visual indicators may correspond to bounding boxes or color overlays. The suspected area 204 provides a visual representation of the precise location of the flame within the FoV 116. The suspected area 204 can be crucial for effective fire management. Users can immediately see where the flame is located, reducing the time required to manually locate the flame and enabling faster intervention.

[0078] Figure 3A An image sequence 202 in a rolling buffer 302 according to an example embodiment of the present disclosure is illustrated. Figure 3B An example embodiment of adding another image from image sequence 202 to scroll buffer 302 is illustrated according to an example embodiment of the present disclosure.

[0079] In some implementations, one or more processors 106 are configured to store image sequences 202 in a rolling buffer 302 in memory at a predefined number of frames per second (FPS). The flame detection system 100 can ensure that this predefined FPS can be continuously recorded and updated, thereby providing rolling recording (location) of FoV. The predefined number of image sequences 202 may correspond to 30. The rolling buffer 302 can hold visual data for 30 seconds at a given time. The image sequences 202 can be sequentially named image 1 to image 30. When a counter reaches 30, indicating that the rolling buffer 302 is full, one or more processors 106 can begin the process of updating the stored image sequences 202.

[0080] In some embodiments, one or more processors 106 are configured to determine whether the flame exists within an image in the image sequence 202 stored in the rolling buffer 302. Furthermore, one or more processors 106 may be configured to save image sequence 202 as a previous frame when it is determined that the flame exists within that image in the image sequence 202 stored in the rolling buffer 302. In some embodiments, image 1 (as shown in 304) is deleted to make room for a new image sequence 202. Image 2 is renamed to image 1. Image 3 is renamed to image 2. The renamed image sequence 202 may continue to be renamed up to image 29. Furthermore, a new image is subsequently captured and stored as image 30. This process ensures that the rolling buffer 302 always contains the most recent image sequence 202.

[0081] In some implementations, upon detecting a flame or alarm event, the flame detection system 100 may take a snapshot of the most current image in the scrolling buffer 302. The most current image in the image sequence 202 may be saved to a folder named "FireMovie1" in memory. This folder may contain image sequences 202 numbered 1 to 30. This folder may save visual data that may have preceded the flame or alarm event. The FireMovie1 may be frozen. The FireMovie1 is stored as a previous frame. After freezing FireMovie1, one or more processors 106 may initiate the storage of a new sequence of image sequences 202. One or more processors 106 may be configured to store another image sequence 202 received from at least one image capture device 104 at a predefined FPS in the scrolling buffer 302 in memory as a current frame. Additionally, a new folder named "FireMovie2" is created. The scrolling buffer 302 process is restarted, and the steps of storing image sequences 202, updating the scrolling buffer 302, and handling new alarm events are repeated. By continuously updating and renaming image sequence 202, flame detection system 100 can use memory efficiently without requiring a large amount of memory (not shown).

[0082] Figure 4 A flowchart 400 illustrating a method for storing fire images in a memory according to an example embodiment of the present disclosure is shown.

[0083] At operation 402, the flame detection system 100 may store the image sequence 202 in a rolling buffer 302 within a folder named FireMovie1. In some embodiments, the flame detection system 100 may initialize the rolling buffer 302 and prepare it for storing the image sequence 202. A folder named "FireMovie1" is created to store the sequence of image sequences 202. The flame detection system 100 may capture the image sequence 202 at regular intervals. The image sequence 202 may be stored sequentially in the rolling buffer 302 within the "FireMovie1" folder. The rolling buffer 302 ensures that the folder always contains the 30 most recent images, thus continuously updating as new image sequences 202 are captured.

[0084] At operation 404, the flame detection system 100 can determine whether a fire alarm has occurred. If no fire alarm has occurred, the flame detection system 100 can store the image sequence 202 in a rolling buffer 302 within a folder named FireMovie1. In some embodiments, the flame detection system 100 can continuously monitor fire alarm events. Monitoring fire alarm events may involve analyzing data from at least one flame detector 102 and an infrared (IR) sensor integrated within at least one flame detector 102. If no fire alarm is detected, the flame detection system 100 can continue to store the image sequence 202 in the rolling buffer 302 within the folder named FireMovie1.

[0085] If a fire alarm occurs, at operation 406, the flame detection system 100 can freeze the folder FireMovie1 and convert the sequence of image sequence 202 into video frames. In some embodiments, upon detecting a fire alarm, the flame detection system 100 can freeze the current state of the "FireMovie1" folder. Freezing the folder corresponds to the current sequence of image sequence 202 being saved, and no additional images in image sequence 202 being written to the folder. The sequence of image sequence 202 stored in the "FireMovie1" folder can be converted into video frames. The video frames can capture moments before and during the fire alarm event.

[0086] At operation 408, the flame detection system 100 may invoke an algorithm to locate regions on a frame. In some embodiments, after converting the image sequence 202 into a video frame, the flame detection system 100 may invoke an algorithm to locate suspicious regions 204 within the video frame. The suspicious region location algorithm may analyze the video frame to identify and highlight areas suspected of being fire sources or potential false alarms. The identified suspicious regions 204 may be marked and highlighted within the video frame.

[0087] Figure 5A The R channel 502, G channel 504 and B channel 506 of each image in the image sequence 202 according to the first embodiment of the present disclosure are illustrated. Figure 5B An example is shown of subtracting one or more pixels of G channel 504 from one or more pixels of R channel 502 according to a first embodiment of the present disclosure. Figure 5C An example of the dilation of image sequence 202 according to a first embodiment of the present disclosure is illustrated. Figure 5D An example of contour extraction of image sequence 202 according to a first embodiment of the present disclosure is illustrated.

[0088] In some implementations, one or more processors 106 are configured to extract red, green, and blue (RGB) frames from previous and current frames. Furthermore, one or more processors 106 may be configured to split the RGB frames of the previous and current frames into R channel 502, G channel 504, and B channel 506. The previous and current frames may be decomposed into a series of RGB frames. Each RGB frame may contain color information in the red channel 502, green channel 504, and blue channel 506. Splitting the RGB frames into R channel 502, G channel 504, and B channel 506 allows for individual analysis of each color channel.

[0089] In some implementations, one or more processors 106 are configured to subtract one or more pixels of the G channel 504 from one or more pixels of the R channel 502. Furthermore, one or more processors 106 may be configured to retain one or more pixels if the difference between the pixel values ​​of the R channel 502 and the G channel 504 is greater than 60. One or more processors 106 may identify potential flame areas based at least on the color intensity difference. Additionally, if the difference (R channel - G channel) 508 exceeds a predefined threshold, one or more pixels may be retained as potential fire pixels. The predefined threshold may correspond to 60.

[0090] In some implementations, one or more processors 106 are configured to dilate the image sequence 202 from previous and current frames by adding one or more pixels of the R channel 502 within the boundary of the flame. One or more processors 106 may add one or more pixels to the boundary of an object to fill holes inside the object. The process of adding one or more pixels to the boundary of an object to fill holes inside the object is called dilation. Dilation can enhance the continuity and integrity of the detected fire area. Dilation can add one or more pixels to the boundary of the object to fill small holes and gaps within the detected fire area. Dilation can help create a more compact and connected area representing fire.

[0091] In some implementations, one or more processors 106 are configured to perform contour extraction on the inflated image sequence 510 by connecting one or more pixels of the R channel 502. One or more processors 106 may define the boundaries of detected fire areas. One or more processors 106 may connect all consecutive white points along the boundaries of the object. Connecting all consecutive white points along the boundaries of the object corresponds to contour extraction on the image sequence. One or more processors 106 may use a contour detection algorithm to connect all consecutive white points along the boundaries of the detected fire areas. Contours (as shown in 512) can be created by tracing the edges of the connected regions, thereby effectively outlining the shape of the fire.

[0092] Figure 6 A Flame 600 is illustrated according to an example implementation of this disclosure.

[0093] In some implementations, the tip Y-coordinate 602 is the maximum height of the coordinates of one or more fire pixels. During contour extraction, each fire pixel in one or more fire pixels within the fire area can be analyzed to find the pixel with the highest y-coordinate. The tip Y-coordinate 602 may represent the highest vertex of the flame 600 to indicate the highest point that the flame 600 can reach. In some implementations, the base Y-coordinate 604 is the coordinate of the lowest pixel of the flame 600. Similarly, during contour extraction, each fire pixel in one or more fire pixels within the fire area can be analyzed to find the pixel with the lowest y-coordinate. The base Y-coordinate 604 may represent the lowest point of the flame 600 to indicate the point where the flame 600 may meet the ground or the base of an object that may be on the flame 600.

[0094] In some implementations, the rightmost X-coordinate 606 is the rightmost pixel coordinate of the flame 600. Contour extraction can identify the pixel with the largest x-coordinate within the fire area. The rightmost X-coordinate 606 can represent the farthest point on the right side of the flame 600 to indicate the extent to which the flame 600 can spread horizontally to the right. The leftmost X-coordinate 608 is the leftmost pixel coordinate of the flame 600. Contour extraction can identify the pixel with the smallest x-coordinate within the fire area. The leftmost X-coordinate 608 can represent the farthest point on the left side of the flame 600 to indicate the extent to which the flame 600 can spread horizontally to the left.

[0095] Figure 7A The base Y coordinate 604, tip Y coordinate 602, leftmost X coordinate 608, and rightmost X coordinate 606 of a plurality of pixels associated with the flame within the FoV in the most current image and previous image in the image sequence 202 according to the first embodiment of the present disclosure are illustrated. Figure 7B Table 706 illustrates predefined limits for determining the height of the flame and predefined limits for the width of the flame 600 according to a first embodiment of the present disclosure.

[0096] In some implementations, the base Y coordinate 604 may represent the lowest point of the flame detected within a frame. In one example, the base Y coordinate 604 of the current frame 704 may correspond to 552 pixels. The base Y coordinate 604 may indicate the lowest point of the flame 600. The tip Y coordinate 602 may represent the highest point of the flame detected within a frame. The tip Y coordinate 602 of the current frame 704 may correspond to 541 pixels. The tip Y coordinate 602 may indicate the highest point of the flame 600. Furthermore, the height of the flame 600 (as shown in 708) may correspond to the base Y coordinate minus the tip Y coordinate. Furthermore, the height of the flame 600 (as shown in 708) may correspond to 552 pixels minus 541 pixels. The height of the flame 600 (as shown in 708) is 11 pixels. The height (as shown in 708) can be derived by subtracting the tip Y coordinate 602 from the base Y coordinate 604. The height (as shown in 708) may represent the vertical span of the flame 600 within a frame. Furthermore, 2% of the flame height (as shown in 708) can correspond to 3 pixels.

[0097] The leftmost X-coordinate 608 corresponds to the minimum x-coordinate of the detected flame 600 within the frame. In the current frame 704, the leftmost X-coordinate 608 corresponds to 987 pixels. The leftmost X-coordinate 606 represents the farthest point to the left of the detectable flame 600. The leftmost X-coordinate 606 marks the starting boundary of the flame 600 on the horizontal axis. The rightmost X-coordinate 606 corresponds to the maximum x-coordinate within the detected fire area. In the current frame 704, the rightmost X-coordinate 606 corresponds to 1002 pixels. The rightmost X-coordinate 606 corresponds to the farthest point to the right of the detectable flame 600. The rightmost X-coordinate 606 marks the ending boundary of the fire area on the horizontal axis. Furthermore, the width of the flame 600 within the frame (as shown in 710) corresponds to the horizontal distance between the leftmost X-coordinate 608 and the rightmost X-coordinate 606. The width (as shown in 710) can be calculated by subtracting the x-coordinate of the leftmost 608 pixels from the x-coordinate of the rightmost 606 pixels. Furthermore, the width of flame 600 (as shown in 710) corresponds to (rightmost x-coordinate - leftmost x-coordinate). Additionally, the width of flame 600 (as shown in 710) corresponds to (1002 pixels - 987 pixels). The width of flame 600 (as shown in 710) is 15 pixels. Furthermore, 2% of the width of flame 600 (as shown in 710) corresponds to 3 pixels.

[0098] In another example, the base Y coordinate 604 of the previous frame 702 could correspond to 554 pixels. Furthermore, the tip Y coordinate 602 of the previous frame 702 could correspond to 536 pixels. Additionally, the leftmost X coordinate 608 could correspond to 993 pixels. Furthermore, the rightmost X coordinate 606 could correspond to 1007 pixels.

[0099] Figure 8A A flowchart 800 illustrating a method for contour extraction of image sequence 202 according to a first embodiment of the present disclosure is shown. Figure 8B A flowchart 800 illustrates a method for storing an image sequence 202 in a memory as a previous frame 702 according to a first embodiment of the present disclosure. Figure 8C A flowchart 800 illustrating a method for contour extraction of another image sequence 202 according to a first embodiment of the present disclosure is shown. Figure 8D A flowchart 800 illustrates a method for storing an image sequence 202 in a memory as the current frame 704 according to a first embodiment of the present disclosure. Figure 8E A flowchart 800 illustrating a method for identifying the location of a flame 600 according to a first embodiment of the present disclosure is shown.

[0100] At operation 802, one or more processors 106 may be configured to extract RGB frames from FireMovie1. In some embodiments, one or more processors 106 may extract RGB frames from a folder named FireMovie1. Extracting RGB frames from a folder named FireMovie1 can initiate the process of identifying flames by obtaining a single frame from a video sequence stored in the FireMovie1 folder. The extracted RGB frames can be used as the basis for image processing and analysis.

[0101] At operation 804, one or more processors 106 can divide an RGB frame into one or more channels. These one or more channels may correspond to R channel 502, G channel 504, and B channel 506, as follows: Figure 5A As shown. Dividing the RGB frame into one or more channels allows for individual analysis of each color channel. One or more colors can provide unique information about the presence and intensity of Flame 600.

[0102] At operation 806, one or more processors 106 may subtract the pixel value of the G channel 504 from the pixel value of the R channel 502, such as Figure 5B As shown. One or more processors 106 can identify potential fire areas based at least on color intensity differences. For each pixel in image sequence 202, the pixel value of green channel 504 is subtracted from the pixel value of red channel 502. In some embodiments, the flame 600 may have a higher intensity in red channel 502 compared to green channel 504. Subtracting the pixel value of red channel 502 from the pixel value of green channel 504 highlights potential fire areas.

[0103] At operation 808, one or more processors 106 may determine whether the difference (R channel pixel value - G channel pixel value) 508 might exceed a predefined threshold. The predefined threshold may correspond to 60. If the difference (R channel pixel value - G channel pixel value) 508 does not exceed the predefined threshold, then at step 802, one or more processors 106 may extract RGB frames from FireMovie1. Determining whether the difference (R channel pixel value - G channel pixel value) 508 exceeds the predefined threshold may be based at least on the color intensity difference to filter out non-fire areas.

[0104] At operation 810, one or more processors 106 may retain pixel values ​​whose difference (R channel pixel value - G channel pixel value) 508 may exceed a predefined threshold. Retaining pixel values ​​whose difference (R channel pixel value - G channel pixel value) 508 exceeds the predefined threshold may isolate pixels that may be part of a fire zone.

[0105] At operation 812, one or more processors 106 may add one or more pixels to the boundary of an object to fill holes inside the object. The process of adding one or more pixels to the boundary of an object to fill holes inside the object is called dilation. Dilation enhances the continuity and integrity of the detected fire area. Dilation can add one or more pixels to the boundary of the object to fill small holes and gaps within the detected fire area. Dilation can help create a tighter and more connected area representing fire.

[0106] At operation 814, one or more processors 106 may define the boundary of the detected fire area. One or more processors 106 may connect all consecutive white points along the boundary of the object. Connecting all consecutive white points along the boundary of the object corresponds to contour extraction. One or more processors 106 may use a contour detection algorithm to connect all consecutive white points along the boundary of the detected fire area. A contour can be created by tracing the edges of the connected areas, thereby effectively outlining the shape of the fire.

[0107] At operation 816, one or more processors 106 may obtain a filtered frame from the previous stage. In some embodiments, the filtered frame may contain only areas that are likely flames 600. The frame may be filtered by thresholding, dilation, and contour extraction. At operation 818, one or more processors 106 may traverse all detected objects in the frame. Traversing all detected objects in the frame ensures that each potential fire area can be analyzed individually. Traversing each object allows system 100 to handle multiple fire areas within a single frame.

[0108] At operation 820, one or more processors 106 can find the extreme points of the object. These extreme points may correspond to the tip Y coordinate 602, the base Y coordinate 604, the rightmost X coordinate 606, and the leftmost X coordinate 608. The tip Y coordinate 602 may correspond to the highest point of the fire zone. The base Y coordinate 604 may correspond to the lowest point of the fire zone. The rightmost X coordinate 606 may correspond to the rightmost point of the fire zone. The leftmost X coordinate 608 may correspond to the leftmost point of the fire zone. The extreme points can define the boundaries of the flame 600 and assess the size of the flame 600.

[0109] At operation 822, one or more processors 106 may store information in memory. Furthermore, one or more processors 106 may mark the memory as "previous memory." In some embodiments, one or more processors 106 may store information about extreme points for each object in memory. This information may be marked "previous" for future reference. Storing this information allows the flame detection system 100 to compare and analyze changes in the fire zone in subsequent frames.

[0110] At operation 824, one or more processors 106 may extract RGB(frame+1) from FireMovie1. In some embodiments, one or more processors 106 may extract RGB(frame+1) from a folder named FireMovie1. Extracting RGB(frame+1) from a folder named FireMovie1 can be initiated by obtaining a single frame from a video sequence stored in the FireMovie1 folder. The extracted RGB(frame+1) can be used as the basis for image processing and analysis.

[0111] At operation 826, one or more processors 106 can divide RGB (frame + 1) into one or more channels. These one or more channels may correspond to R channel 502, G channel 504, and B channel 506. Dividing RGB (frame + 1) into these one or more channels allows for individual analysis of each color channel. One or more colors can provide unique information about the presence and intensity of the flame 600.

[0112] At operation 828, one or more processors 106 may subtract the pixel value of the G channel 504 from the pixel value of the R channel 502. The one or more processors 106 may identify potential fire areas based at least on the color intensity difference. For each pixel in image sequence 202, the pixel value of the green channel 504 is subtracted from the pixel value of the red channel 502. In some embodiments, the flame 600 may have a higher intensity in the red channel 502 compared to the green channel 504. Subtracting the pixel value of the red channel 502 from the pixel value of the green channel 504 highlights potential fire areas.

[0113] At operation 830, it is determined whether the difference (R channel pixel value - G channel pixel value) 508 might exceed a predefined threshold. The predefined threshold may correspond to 60. If the difference (R channel pixel value - G channel pixel value) 508 might not exceed the predefined threshold, then at step 824, one or more processors 106 may extract RGB (frame + 1) from FireMovie1. Determining whether the difference (R channel pixel value - G channel pixel value) 508 might exceed the predefined threshold may be based at least on the color intensity difference to filter out non-fire areas.

[0114] At operation 832, pixel values ​​whose difference (R channel pixel value - G channel pixel value) 508 may exceed a predefined threshold are retained. Retaining pixel values ​​whose difference (R channel pixel value - G channel pixel value) 508 may exceed a predefined threshold can isolate pixels that may be part of a fire zone.

[0115] At operation 834, one or more processors 106 may add one or more pixels to the boundary of an object to fill holes inside the object. The process of adding one or more pixels to the boundary of an object to fill holes inside the object is called dilation. Dilation enhances the continuity and integrity of the detected fire area. Dilation can add one or more pixels to the boundary of the object to fill small holes and gaps within the detected fire area. Dilation can help create a tighter and more connected area representing fire.

[0116] At operation 836, one or more processors 106 may define the boundary of the detected fire area. One or more processors 106 may connect all consecutive white points along the boundary of the object. Connecting all consecutive white points along the boundary of the object corresponds to contour extraction. One or more processors 106 may use a contour detection algorithm to connect all consecutive white points along the boundary of the detected fire area. A contour can be created by tracing the edges of the connected areas, thereby effectively outlining the shape of the fire.

[0117] At operation 838, one or more processors 106 may obtain a filtered frame from the previous stage. In some embodiments, the filtered frame may contain only areas that are likely flames 600. The frame may be filtered by thresholding, dilation, and contour extraction. At operation 840, one or more processors 106 may traverse all detected objects in the frame. Traversing all detected objects in the frame ensures that each potential fire area can be analyzed individually. Traversing each object allows the flame detection system 100 to handle multiple fire areas within a single frame.

[0118] At operation 842, one or more processors 106 can find the extreme points of the object. These extreme points may correspond to the tip Y coordinate 602, the base Y coordinate 604, the rightmost X coordinate 606, and the leftmost X coordinate 608. The tip Y coordinate 602 may correspond to the highest point of the fire area. The base Y coordinate 604 may correspond to the lowest point of the fire area. The right X coordinate 606 may correspond to the rightmost point of the fire area. The left X coordinate 608 may correspond to the leftmost point of the fire area. The extreme points can define the boundaries of the flame 600 and assess the size of the flame 600.

[0119] At operation 844, one or more processors 106 may store information in memory. Furthermore, one or more processors 106 may mark the memory as current memory. In some embodiments, one or more processors 106 may store information about extreme points for each object in memory. This information may be marked as "current" for comparison. Storing this information allows the flame detection system 100 to compare and analyze changes in the fire zone in subsequent frames.

[0120] At operation 846, one or more processors 106 may pick up two memory data sets, "current" and "previous". Comparing the "current" and "previous" data enables the flame detection system 100 to detect changes in the properties of a fire zone over time. At operation 848, one or more processors 106 may compare the base Y coordinates of one or more fire pixels from the previous and current times to determine whether their changes are less than 2% of the flame height. This comparison checks whether the change is less than 2% of the flame height.

[0121] At operation 850, one or more processors 106 may compare the tip Y-coordinates of one or more fire pixels from previous and current data to determine if their changes are greater than 2% of the flame height. In some embodiments, one or more processors 106 may compare the tip Y-coordinates of one or more fire pixels from "current" and "current" data. This comparison may check if the change is greater than 2% of the flame height.

[0122] At operation 852, one or more processors 106 may compare the leftmost X-coordinates of one or more fire pixels from previous and current data to determine whether their changes are greater than 2% of the flame width. In some embodiments, one or more processors 106 may compare the leftmost X-coordinates of one or more fire pixels from "current" and "current" data. This comparison may check whether the change is greater than 2% of the flame width.

[0123] At operation 854, one or more processors 106 may compare the rightmost X coordinates of one or more fire pixels from previous and current data to determine if their changes are greater than 2% of the flame width. In some embodiments, system 100 may compare the rightmost X coordinates of one or more fire pixels from "current" and "previous" data. This comparison may check if the change is greater than 2% of the flame width. At operation 856, one or more processors 106 may retain only objects on the frame that meet the comparison criteria. This comparison may filter out irrelevant changes in the fire area.

[0124] The presence of a flickering flame can be indicated by determining that the base Y-coordinate of one or more fire pixels changes by less than 2%, but at least one of the tip Y-coordinate, leftmost X-coordinate, or rightmost X-coordinate changes by more than 2%. For example, the flame is not a stationary object, but a flickering one. Thus, the base Y-coordinate of the flame may not change by more than a threshold frame by frame, but the tip, as well as the left and right edges, may change by more than a threshold frame by frame.

[0125] Figure 9A An example is illustrated of a flame 600 having a boundary 902 around the flame 600 according to a second embodiment of the present disclosure. Figure 9B The illustration shows a vertical line 906 between the tip Y coordinate 602 and the base Y coordinate 604 of the flame 600 according to a second embodiment of the present disclosure, and a horizontal line 908 between the leftmost X coordinate 608 and the rightmost X coordinate 606 of the flame 600. Figure 9C A plurality of pixels 910 associated with flame 600 are illustrated according to a second embodiment of the present disclosure.

[0126] In some embodiments, the flame detection system 100 is configured to determine the position of the flame 600 within the FoV 116. In some embodiments, one or more processors 106 are configured to determine a vertical line 906 between the tip Y-coordinate 602 and the base Y-coordinate 604 of the flame 600. Furthermore, one or more processors 106 may be configured to determine a horizontal line 908 between the leftmost X-coordinate 608 and the rightmost X-coordinate 606 of the flame 600. Additionally, one or more processors 106 may be configured to determine the intersection of the vertical line 906 and the horizontal line 908. Furthermore, one or more processors 106 may be configured to associate this intersection with a center pixel 912.

[0127] In some embodiments, one or more processors 106 are configured to determine the midpoint (M1) coordinate 904 between the tip Y-coordinate 602 and the base Y-coordinate 604 of the flame source from the previous frame. Furthermore, one or more processors 106 may be configured to determine the midpoint (M2) coordinate 904 between the left X-coordinate 608 and the right X-coordinate 606 of the flame source from the previous frame. Additionally, one or more processors 106 may be configured to determine that the midpoint (M1) coordinate is equal to the midpoint (M2) coordinate. Furthermore, one or more processors 106 may be configured to determine that the midpoint (M1) coordinate 904 and the midpoint (M2) coordinate 904 are labeled as the coordinates of the center pixel 912 for the previous frame.

[0128] In some embodiments, the center of flame 600 is calculated as the midpoint 904 of the boundary formed by the extreme points of flame 600. One or more processors 106 may determine the midpoint 904 on each edge of the formed boundary 902. One or more processors 106 may also mark the determined midpoint 904. Furthermore, one or more processors 106 may connect the midpoints 904 on the formed boundary 902. In addition, the point intersecting with the midpoint 904 may be marked as the center of flame 600.

[0129] In some implementations, one or more processors 106 may verify that the intensity of one or more pixels 910 at the center pixel 912 is equal to the intensity of adjacent pixels 914. One or more processors 106 may extract the pixel intensity values ​​of the center pixel 912 and its directly adjacent pixels 914. One or more processors 106 may compare the intensity of the center pixel 912 with that of its adjacent pixels 914. The adjacent pixels 914 may include pixels located directly above, below, to the left, and to the right of the center pixel 912. The center pixel 912 and adjacent pixels 914 may be represented in a grid format, thus representing the flame 600 in rows and columns.

[0130] Figure 10A An image sequence 202 is illustrated, showing that the intensity value of the center pixel 912 according to a second embodiment of the present disclosure is equal to the intensity of each of the adjacent pixels 914. Figure 10B Table 1012 illustrates an image sequence 202 according to a second embodiment of the present disclosure, showing that the intensity value of the center pixel 912 is equal to the intensity of each of the adjacent pixels 914.

[0131] In some embodiments, one or more processors 106 may verify that the intensity of the center pixel 912 of the flame is nearly identical to the intensity of its neighboring pixels 914. Image sequence 202 may have a center pixel 912 of the flame at X coordinate = 1000 and Y coordinate = 545. The center pixel 912 of the flame may have an RGB intensity of [211, 74, 0] (as shown in 1002). Furthermore, image sequence 202 may have neighboring pixels 914 of the flame at X coordinate = 999 and Y coordinate = 545 (located to the left of the center). The neighboring pixels 914 of the flame may have an RGB intensity of [211, 74, 0] (as shown in 1004). Additionally, image sequence 202 may have neighboring pixels 914 of the flame at X coordinate = 1001 and Y coordinate = 545 (located to the right of the center). The neighboring pixels 914 of the flame may have an RGB intensity of [211, 74, 0] (as shown in 1006). Furthermore, image sequence 202 may have adjacent pixels 914 of the flame at X coordinate = 1000 and Y coordinate = 544 (located above the center). The adjacent pixels 914 of the flame may have RGB intensities [211, 74, 0] (as shown in 1008). Subsequently, image sequence 202 may have adjacent pixels 914 of the flame at X coordinate = 1000 and Y coordinate = 546 (located below the center). The adjacent pixels 914 of the flame may have RGB intensities [211, 74, 0] (as shown in 1010).

[0132] In some implementations, the verification process confirms that the center pixel 912 of the flame at coordinates (1000, 545) has an RGB intensity of [211, 74, 0] and is consistent with its directly adjacent pixel 914. Each adjacent pixel 914 of the flame (left, right, top, and bottom) may have the same RGB intensity.

[0133] Figure 11A A flowchart 1100 illustrating a method for contour extraction of image sequence 202 according to a second embodiment of the present disclosure is shown. Figure 11B A flowchart 1100 illustrates a method for determining the center pixel 912 of a plurality of pixels 910 associated with the flame 600 within the FoV 116 for at least one image in an image sequence 202, according to a second embodiment of the present disclosure. Figure 11C A flowchart 1100 illustrating a method for identifying the location of a flame 600 according to a second embodiment of the present disclosure is shown.

[0134] At operation 1102, one or more processors 106 may extract RGB frames from FireMovie1. In some embodiments, one or more processors 106 may extract RGB frames from a folder named FireMovie1. Extracting RGB frames from a folder named FireMovie1 can be initiated by obtaining a single frame from a video sequence stored in the FireMovie1 folder. The extracted RGB frames can be used as the basis for image processing and analysis.

[0135] At operation 1104, one or more processors 106 can divide an RGB frame into one or more channels. These one or more channels may correspond to R channel 502, G channel 504, and B channel 506. Dividing the RGB frame into these one or more channels allows for individual analysis of each color channel. One or more colors can provide unique information about the presence and intensity of the flame 600.

[0136] At operation 1106, one or more processors 106 may subtract the pixel value of the G channel 504 from the pixel value of the R channel 502. The one or more processors 106 may identify potential fire areas based at least on the color intensity difference. For each pixel in image sequence 202, the pixel value of the green channel 504 is subtracted from the pixel value of the red channel 502. In some embodiments, the flame 600 may have a higher intensity in the red channel 502 compared to the green channel 504. Subtracting the pixel value of the red channel 502 from the pixel value of the green channel 504 highlights potential fire areas.

[0137] At operation 1108, it is determined whether the difference (R channel pixel value - G channel pixel value) 508 might exceed a predefined threshold. The predefined threshold may correspond to 60. If the difference (R channel pixel value - G channel pixel value) 508 might not exceed the predefined threshold, then at step 1102, one or more processors 106 may extract RGB frames from FireMovie1. Determining whether the difference (R channel pixel value - G channel pixel value) 508 might exceed the predefined threshold may be based at least on the color intensity difference to filter out non-fire areas.

[0138] At operation 1110, pixel values ​​whose difference (R channel pixel value - G channel pixel value) 508 may exceed a predefined threshold are retained. Retaining pixel values ​​whose difference (R channel pixel value - G channel pixel value) 508 may exceed a predefined threshold can isolate pixels that may be part of a fire zone.

[0139] At operation 1112, one or more processors 106 may add one or more pixels to the boundary of an object to fill holes inside the object. The process of adding one or more pixels to the boundary of an object to fill holes inside the object is called dilation. Dilation enhances the continuity and integrity of the detected fire area. Dilation can add one or more pixels to the boundary of the object to fill small holes and gaps within the detected fire area. Dilation can help create a tighter and more connected area representing fire.

[0140] At operation 1114, one or more processors 106 may define the boundary of the detected fire area. One or more processors 106 may connect all consecutive white points along the boundary of the object. Connecting all consecutive white points along the boundary of the object corresponds to contour extraction. One or more processors 106 may use a contour detection algorithm to connect all consecutive white points along the boundary of the detected fire area. By tracing the edges of the connected areas, a contour is created, thereby effectively outlining the shape of the fire.

[0141] At operation 1116, one or more processors 106 can find the extreme points of an object within a frame. These extreme points may correspond to the tip Y coordinate 602, the base Y coordinate 604, the rightmost X coordinate 606, and the leftmost X coordinate 608. The tip Y coordinate 602 may correspond to the highest point of the fire zone. The base Y coordinate 604 may correspond to the lowest point of the fire zone. The rightmost X coordinate 606 may correspond to the rightmost point of the fire zone. The leftmost X coordinate 608 may correspond to the leftmost point of the fire zone. The extreme points may define the boundaries of the flame 600 and assess the size of the flame 600.

[0142] At operation 1118, one or more processors 106 may determine a midpoint 1 between the tip Y-coordinate 602 and the base Y-coordinate 604 of the flame 600. Midpoint 1 may help determine the vertical center of the flame 600. Midpoint 1 may provide an average position between the tip Y-coordinate 602 and the base Y-coordinate 604.

[0143] At operation 1120, one or more processors 106 may determine the midpoint 2 between the leftmost X coordinate 608 and the rightmost X coordinate 606 of the flame 600. Midpoint 2 may help determine the horizontal center of the flame 600. Midpoint 2 may provide an average position between the right X coordinate 606 and the left X coordinate 608.

[0144] At operation 1122, one or more processors 106 may determine whether the coordinates (X,Y)904 of midpoint 1 are equal to the coordinates (X,Y)904 of midpoint 2. If the coordinates (X,Y)904 of midpoint 1 may not be equal to the coordinates (X,Y)904 of midpoint 2, one or more processors 106 may find the extreme point of the object in the frame (as shown at step 1116). This comparison ensures the accuracy of the detected center of the flame 600. If the coordinates (X,Y)904 of midpoint 1 may not match the coordinates (X,Y)904 of midpoint 2, it indicates that the center may not be correctly identified. The leftmost X coordinate 608 may redetermine the extreme point and the center. At operation 1124, if the coordinates (X,Y)904 of midpoint 1 match the coordinates (X,Y)904 of midpoint 2, one or more processors 106 may mark the coordinates as the center (X,Y).

[0145] At operation 1126, one or more processors 106 may determine whether the pixel RGB intensity of the object at the center (X,Y) pixel 912 is equal to the intensity (X,{Y+1}) of the adjacent RGB pixel 914. Furthermore, one or more processors 106 may determine whether the pixel RGB intensity of the object at the center (X,Y) pixel 912 is equal to the intensity (X,{Y-1}) of the adjacent RGB pixel 912. Furthermore, one or more processors 106 may determine whether the pixel RGB intensity of the object at the center (X,Y) pixel 912 is equal to the intensity ({X+1},Y) of the adjacent RGB pixel 914. Furthermore, one or more processors 106 may determine whether the pixel RGB intensity of the object at the center (X,Y) pixel 912 is equal to the intensity ({X-1},Y) of the adjacent RGB pixel 914.

[0146] At operation 1128, one or more processors 106 may retain objects on the frame that meet the intensity consistency criteria. In some embodiments, one or more processors 106 may filter out objects that may not exhibit the same intensity. One or more processors 106 may ensure that only valid fire zones are retained. At operation 1130, one or more processors 106 may determine whether one or more processors 106 can traverse all objects in the frame. In some embodiments, one or more processors 106 may ensure that the frame (along with all detected objects) is thoroughly analyzed.

[0147] Figure 12 An example is illustrated by aligning the flashing frequency of at least one flame detector 102 with the flashing frequency of an image sequence 202 captured by at least one image capture device 104 according to a third embodiment of the present disclosure. Figure 13Multiple graphs illustrating the flashing frequency of at least one flame detector 102 according to a third embodiment of the present disclosure and the flashing frequency of an image sequence 202 captured by at least one image capturing device 104 are shown.

[0148] In some implementations, to ensure that the flicker frequency of the IR sensor 102 (which also corresponds to at least one flame detector 102) is aligned with the flicker frequency of at least one image capture device 104, the flame detection system 100 may capture data at a specified rate, apply Fourier analysis to transform the data from the time domain to the frequency domain, and then the flame detection system 100 may compare the results to ensure synchronization. In some implementations, the IR sensor 102 data is captured at a rate of 60 samples per second (60 Hz). The Fourier analysis may correspond to a Fast Fourier Transform (FFT).

[0149] Fourier analysis can be applied to the captured data from the IR sensor 102. Fourier analysis transforms the signal from the original time domain to the frequency domain. In the time domain, the signal can be represented by the change in amplitude over time. The FFT transforms the time domain to the frequency domain. In the frequency domain, the signal can be represented by frequency and amplitude. The flame detection system 100 can calculate the FFT of the captured data, thereby generating a spectrum. Each frequency can represent a possible flicker frequency of the IR sensor 102. In some embodiments, the flame detection system 100 can identify peak frequencies in the spectrum obtained from the FFT. The peak frequencies can correspond to the dominant flicker frequency of the IR sensor 102.

[0150] In some embodiments, the flame detection system 100 may perform frequency analysis on data from at least one image capture device 104 to determine the flicker frequency. The dominant frequency from at least one image capture device 104 may be compared with the dominant frequency obtained from an IR sensor 102.

[0151] In one example, graph 1300 may represent the flicker frequency of the IR sensor 102 in a previous frame. Graph 1300 may correspond to a graph between PSD (as shown in 1302) and frequency (as shown in 1304). Furthermore, graph 1306 may represent the flicker frequency of at least one image capture device 104 in a previous frame. Graph 1306 may correspond to a graph between the average value of pixels within the fire outline (as shown in 1308) and frequency (as shown in 1310). The flicker frequencies of the IR sensor 102 and at least one image capture device 104 may be aligned at 3 Hz and 11 Hz frequencies.

[0152] In another example, graph 1312 may represent the flicker frequency of the IR sensor 102 in the current frame. Graph 1312 may correspond to a graph between PSD (as shown in 1302) and frequency (as shown in 1304). Furthermore, graph 1318 may represent the flicker frequency of at least one image capture device 104 in a previous frame. Graph 1318 may correspond to a graph between the average value of pixels within the fire outline (as shown in 1320) and frequency (as shown in 1322). The flicker frequencies of the IR sensor 102 and at least one image capture device 104 may be aligned to 3 Hz and 11 Hz frequencies.

[0153] Figure 14A A flowchart 1400 illustrating a method for contour extraction of image sequence 202 according to a third embodiment of the present disclosure is shown. Figure 14B A flowchart 1400 illustrates the steps for determining the flashing frequency of at least one flame detector 102 and the flashing frequency of an image sequence 202 according to a third embodiment of the present invention. Figure 14C A flowchart 1400 illustrating a method for identifying the location of a flame 600 according to a third embodiment of the present disclosure is shown.

[0154] At operation 1402, the flame detection system 100 may extract RGB frames from FireMovie1. In some embodiments, the flame detection system 100 may extract RGB frames from a folder named FireMovie1. Extracting RGB frames from a folder named FireMovie1 can be initiated by obtaining a single frame from a video sequence stored in the FireMovie1 folder. The extracted RGB frames can be used as the basis for image processing and analysis.

[0155] At operation 1404, the flame detection system 100 can divide an RGB frame into one or more channels. These one or more channels may correspond to R channel 502, G channel 504, and B channel 506. Dividing the RGB frame into these one or more channels allows for individual analysis of each color channel. One or more colors can provide unique information about the presence and intensity of the flame 600.

[0156] At operation 1406, the flame detection system 100 may subtract the pixel value of the G channel 504 from the pixel value of the R channel 502. The flame detection system 100 may identify potential fire areas based at least on the color intensity difference. For each pixel in the image sequence 202, the pixel value of the green channel 504 is subtracted from the pixel value of the red channel 502. In some embodiments, the flame 600 may have a higher intensity in the red channel 502 compared to the green channel 504. Subtracting the pixel value of the red channel 502 from the pixel value of the green channel 504 highlights potential fire areas.

[0157] At operation 1408, it is determined whether the difference (R channel pixel value - G channel pixel value) 508 might exceed a predefined threshold. The predefined threshold may correspond to 60. If the difference (R channel pixel value - G channel pixel value) 508 might not exceed the predefined threshold, then at step 1402, the flame detection system 100 may extract RGB frames from FireMovie1. Determining whether the difference (R channel pixel value - G channel pixel value) 508 might exceed the predefined threshold may be based at least on the color intensity difference to filter out non-fire areas.

[0158] At operation 1410, pixel values ​​whose difference (R channel pixel value - G channel pixel value) 508 may exceed a predefined threshold are retained. Retaining pixel values ​​whose difference (R channel pixel value - G channel pixel value) 508 may exceed a predefined threshold can isolate pixels that may be part of a fire zone.

[0159] At operation 1412, the flame detection system 100 may add one or more pixels to the boundary of an object to fill holes inside the object. The process of adding one or more pixels to the boundary of an object to fill holes inside the object is called dilation. Dilation enhances the continuity and integrity of the detected fire area. Dilation may add one or more pixels to the boundary of the object to fill small holes and gaps within the detected fire area. Dilation helps create a tighter and more connected area representing fire.

[0160] At operation 1414, the flame detection system 100 can define the boundary of the detected fire area. The flame detection system 100 can connect all consecutive white dots along the boundary of the object. Connecting all consecutive white dots along the boundary of the object corresponds to contour extraction. The flame detection system 100 can use a contour detection algorithm to connect all consecutive white dots along the boundary of the detected fire area. A contour can be created by tracing the edges of the connected areas, thereby effectively outlining the shape of the fire.

[0161] At operation 1416, the flame detection system 100 can find a single average value for each individual object in the frame. In some embodiments, for each object in the frame, the flame detection system 100 can calculate an average pixel intensity. The average value can represent the brightness or intensity of the object. At operation 1418, the flame detection system 100 can iterate through each object in the frame and store the average value in memory. The flame detection system 100 can calculate the average value for each object and store it in memory. The stored average value can be used for time-domain to frequency-domain conversion.

[0162] At operation 1420, the flame detection system 100 can determine whether it can traverse 30 consecutive frames. Traversing 30 frames provides a dataset for frequency analysis. Frequency analysis can capture potential changes over time.

[0163] At operation 1422, the flame detection system 100 can convert a time-domain average value from memory into a frequency-domain representation. In some embodiments, the flame detection system 100 can apply a Fourier transform to the stored average value. The transformation can identify the dominant frequency of intensity variation. The transformation can also represent the flicker frequency of the object.

[0164] At operation 1424, the flame detection system 100 can determine whether the frequencies of the IR sensor 102 and at least one image capture device 104 are aligned. In some embodiments, the flame detection system 100 can compare frequency domain data obtained from the IR sensor 102 and at least one image capture device 104. The flame detection system 100 can also check whether the dominant frequencies match within a specified tolerance range. At operation 1426, the flame detection system 100 can retain aligned frequency-preserving objects. The flame detection system 100 can filter out objects whose flicker frequencies are not aligned with the flicker frequency of the IR sensor 102.

[0165] This disclosure offers a significant advantage by integrating at least one image capture device 104 with at least one flame detector 102, enabling precise location of flame sources. This integration allows for real-time image capture and processing, enabling the exact location of the flame 600 to be directly marked on the image sequence 202. Precise identification of the flame 600's location greatly assists operators in quickly and accurately locating the flame source, thereby enhancing safety and efficiency during emergencies. This disclosure also reduces false alarms and improves response time.

[0166] Those skilled in the art to which this invention pertains will, upon benefiting from the teachings presented in the foregoing description and accompanying drawings, contemplate numerous modifications and other embodiments of the invention set forth herein. Therefore, it should be understood that the invention is not limited to the specific embodiments disclosed, and that modifications and other embodiments are intended to be included within the scope of the appended claims. Furthermore, although the foregoing description and accompanying drawings have described exemplary embodiments in the context of certain example combinations of elements and / or functions, it should be understood that different combinations of elements and / or functions may be provided by alternative embodiments without departing from the scope of the appended claims. In this regard, for example, different combinations of elements and / or functions explicitly described above are also contemplated, as may be set forth in some of the appended claims. Although specific terms are used herein, they are used only in a general and descriptive sense and not for limiting purposes.

Claims

1. A flame detection system comprising: at least one flame detector configured to generate data corresponding to detection of a flame being present within a field of view (FoV); at least one image capture device operatively coupled to the at least one flame detector, wherein the at least one image capture device is configured to capture a sequence of images of the flame within the FoV, wherein a plurality of pixels of each image in the sequence of images are associated with the flame within the FoV; and one or more processors communicatively coupled to the at least one flame detector and the at least one image capture device, wherein the one or more processors are configured to identify a location of the flame within the FoV by: determining, for each image in the sequence of images, a base Y coordinate, a tip Y coordinate, a leftmost X coordinate, and a rightmost X coordinate of the plurality of pixels associated with the flame within the FoV; determining that a change in the base Y coordinate of a most current image in the sequence of images compared to a previous image in the sequence of images is less than a predefined limit of a height of the flame; and determining that a change in the tip Y coordinate of the most current image in the sequence of images compared to the previous image in the sequence of images is greater than the predefined limit of the height of the flame, or a change in the leftmost X coordinate or the rightmost X coordinate of the most current image in the sequence of images compared to the previous image in the sequence of images is greater than a predefined limit of a width of the flame.

2. The flame detection system of claim 1, wherein the predefined limit of the height of the flame corresponds to 2% of the height of the flame, and the predefined limit of the width of the flame corresponds to 2% of the width of the flame.

3. The flame detection system of claim 1, wherein the one or more processors are further configured to: store the sequence of images in a rolling buffer within a memory at a predefined frames per second (FPS); determine whether the flame is present within an image in the sequence of images stored in the rolling buffer; upon determining that the flame is present within the image in the sequence of images stored in the rolling buffer within the memory, save the image in the sequence of images as a previous frame; and store another sequence of images received from the at least one image capture device in the rolling buffer within the memory as a current frame at the predefined FPS.

4. The flame detection system of claim 3, wherein the predefined FPS defines a range between 30 FPS and 60 FPS.

5. The flame detection system of claim 3, wherein the one or more processors are configured to determine the base Y coordinate, the tip Y coordinate, the leftmost X coordinate, and the rightmost X coordinate by: extracting a red-green-blue (RGB) frame from the previous image and the most current image in the sequence of images; dividing the RGB frame of the previous image and the most current image into an R channel, a G channel, and a B channel; subtracting one or more pixels of the G channel from one or more pixels of the R channel; determining that a difference in pixel values between the R channel and the G channel is greater than 60; dilating the sequence of images from the previous frame and the current frame by adding the one or more pixels of the R channel within a boundary of the flame; and contouring the dilated sequence of images by connecting the one or more pixels of the R channel.

6. The flame detection system of claim 3, wherein the plurality of pixels of the sequence of images correspond to one or more pixels of the current frame and the one or more pixels of the previous frame.

7. The flame detection system of claim 1, wherein the data corresponds to infrared (IR) sensor data captured at a rate of 60 samples per second.

8. The flame detection system of claim 1, wherein the flame detection system is further configured to indicate the location of the flame within the FoV on at least one image in the sequence of images.

9. A flame detection system, the flame detection system comprising: at least one flame detector configured to generate data corresponding to detecting a presence of a flame within a field of view (FoV); at least one image capture device operatively coupled to the at least one flame detector, wherein the at least one image capture device is configured to capture a sequence of images of the flame within the FoV, wherein a plurality of pixels of each image in the sequence of images are associated with the flame within the FoV; and one or more processors communicatively coupled to the at least one flame detector and the at least one image capture device, wherein the one or more processors are configured to identify a location of the flame within the FoV by: determining, for at least one image in the sequence of images, a center pixel of the plurality of pixels associated with the flame within the FoV; determining an intensity of the center pixel and an intensity of neighboring pixels, wherein the neighboring pixels include a pixel to the left of the center pixel, a pixel to the right of the center pixel, a pixel directly below the center pixel, and a pixel directly above the center pixel; and determining that an intensity value of the center pixel is equal to the intensity of each of the neighboring pixels.

10. The flame detection system of claim 9, wherein the one or more processors are further configured to determine, for the at least one image in the sequence of images, coordinates of the center pixel of the plurality of pixels associated with the flame within the FoV by: determining a perpendicular line between a tip Y coordinate and a base Y coordinate of the flame; determining a horizontal line between a leftmost X-coordinate and a rightmost X-coordinate of the flame; determining an intersection of the vertical line and the horizontal line; and associating the intersection as the center pixel.

Images