Flame detection device, flame detection method, and program

Abstract

To detect flame with high accuracy using a method different from a determination with a circularity.SOLUTION: A flame detection apparatus includes: an acquisition unit that acquires a target image in which a monitored target is captured; an extraction unit that extracts a flame candidate region from the target image acquired by the acquisition unit; and a determination unit that calculates an index indicating a complexity of a shape in the flame candidate region by using a statistic amount of an angle forming a contour line connecting a pixel group constituting a contour of the flame candidate region, and determines whether or not flame is indicated in the flame candidate region based on the calculated index.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a flame detection device, a flame detection method, and a program. [Background technology]

[0002] Techniques for detecting the occurrence of a fire using images are known (for example, Patent Documents 1 and 2). A determination is made for each pixel in an image as to whether or not a flame-like object is shown, and an area consisting of a collection of flame-like pixels is extracted. If the shape of the extracted area is far from circular, it is determined that the area shows a flame. For example, circularity (= 4πS / L^2) can be used as an index of the degree of circularity. Here, S indicates the area of ​​the area, and L indicates the perimeter of the area. By excluding nearly circular objects from the flame picture, it is possible to prevent luminous objects such as rotating lights from being mistakenly detected as a flame, thereby improving detection accuracy. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 8-305980 [Patent Document 2] Japanese Patent Publication No. 2020-021300 Summary of the Invention [Problem to be solved by the invention]

[0004] However, the determination based on circularity does not necessarily allow accurate detection of flames. For example, some light-emitting objects, such as fluorescent lamps, have a shape that is close to a rectangle. In an image capturing the light emitted from such a rectangular light-emitting object, the shape of the light-emitting area is far from circular, and there is a risk that it will be mistakenly detected as a flame. Naturally, when detecting a flame, it is desirable not to mistakenly detect a light-emitting object that has a shape other than a circle, such as a rectangle, as a flame.

[0005] The present invention has been made in consideration of the above circumstances, and its purpose is to provide a flame detection device, a flame detection method, and a program that can accurately detect flames using a method other than determination based on circularity. [Means for solving the problem]

[0006] In order to solve the above-mentioned problems, one embodiment of the present invention is a flame detection device that includes an acquisition unit that acquires a target image of a monitored object, an extraction unit that extracts a flame candidate region from the target image acquired by the acquisition unit, and a determination unit that calculates an index indicating the complexity of the shape in the flame candidate region using statistics of the angle formed by a contour line connecting groups of pixels that make up the contour of the flame candidate region, and determines whether a flame is shown in the flame candidate region based on the calculated index.

[0007] Moreover, one embodiment of the present invention is a flame detection method performed by a flame detection device that is a computer, in which an acquisition unit acquires a target image of a monitored object, an extraction unit extracts a flame candidate region from the target image acquired by the acquisition unit, and a determination unit calculates an index indicating the complexity of the shape in the flame candidate region using statistics of the angle formed by a contour line connecting groups of pixels that make up the contour of the flame candidate region, and determines whether a flame is indicated in the flame candidate region based on the calculated index.

[0008] Moreover, one embodiment of the present invention is a program that acquires a target image of a monitored object, extracts a flame candidate region from the acquired target image, calculates an index indicating the complexity of the shape of the flame candidate region using statistics of the angle formed by a contour line connecting groups of pixels that make up the outline of the flame candidate region, and determines whether or not a flame is shown in the flame candidate region based on the calculated index. [Effects of the Invention]

[0009] As described above, according to the present invention, it is possible to accurately detect a flame using a method different from the determination based on circularity. [Brief explanation of the drawings]

[0010] [Figure 1] 1 is a diagram showing an example of the configuration of a flame detection system 1 to which a flame detection device 10 according to an embodiment is applied. [Figure 2] 1 is a block diagram showing an example of the configuration of a flame detection device 10 according to an embodiment. [Figure 3A] 3 is a diagram illustrating a process performed by a flame determination unit 102 according to the embodiment. FIG. [Figure 3B] 3 is a diagram illustrating a process performed by a flame determination unit 102 according to the embodiment. FIG. [Figure 3C] 3 is a diagram illustrating a process performed by a flame determination unit 102 according to the embodiment. FIG. [Figure 4] 3 is a flowchart showing the flow of processing performed by the flame detection device 10 according to the embodiment. [Figure 5] 3 is a flowchart showing the flow of processing performed by the flame detection device 10 according to the embodiment. [Figure 6] 3 is a flowchart showing the flow of processing performed by the flame detection device 10 according to the embodiment. [Figure 7] 1A to 1C are diagrams illustrating examples of a target image, a flame candidate region, statistics, and indices according to an embodiment. [Figure 8] 1A to 1C are diagrams illustrating examples of a target image, a flame candidate region, statistics, and indices according to an embodiment. [Figure 9] 1A to 1C are diagrams illustrating examples of a target image, a flame candidate region, statistics, and indices according to an embodiment. [Figure 10] 1A to 1C are diagrams illustrating examples of a target image, a flame candidate region, statistics, and indices according to an embodiment. [Figure 11] 10A and 10B are diagrams illustrating examples of indices when a target image is a moving image according to an embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0012] 1 is a diagram showing an example configuration of a flame detection system 1 to which a flame detection device 10 according to an embodiment is applied. The flame detection system 1 includes, for example, a camera CA and the flame detection device 10. The camera CA and the flame detection device 10 are communicatively connected via a communication network, short-range communication using a wireless LAN (Local Area Network), or the like, or via a USB (Universal Serial Bus) cable or the like. The monitored object T is an area or object that is the target of monitoring by the flame detection system 1. The monitored object T may be set arbitrarily. For example, the monitored object T may be a roadway, a tunnel premises, a building, a building premises, etc. The camera CA captures an image of the monitoring target T. The camera CA may capture a moving image or a still image. The camera CA outputs the captured image to the flame detection device 10. Flame detection device 10 is a computer, and is realized by, for example, a PC (Personal Computer), a server device, etc. Flame detection device 10 acquires an image (target image) of a monitoring target T from a camera CA, and determines whether a flame is shown in the acquired image.

[0013] 2 is a block diagram showing an example configuration of flame detection device 10 according to an embodiment. Flame detection device 10 includes, for example, image acquisition unit 100, flame candidate region extraction unit 101, flame determination unit 102, determination result output unit 103, and image information storage unit 104. The functional units of the flame detection device 10 (image acquisition unit 100, flame candidate area extraction unit 101, flame judgment unit 102, and judgment result output unit 103) are realized by having a CPU (Central Processing Unit) that the flame detection device 10 has as hardware execute a program.

[0014] The image acquisition unit 100 acquires image information of a target image. The image acquisition unit 100 acquires image information of a target image captured by a camera CA. The image acquisition unit 100 outputs the acquired image information to the flame candidate region extraction unit 101.

[0015] The flame candidate region extraction unit 101 extracts a flame candidate region from a target image. A flame candidate region is a region in the target image that shows an object that looks like a flame. The flame candidate region extraction unit 101 extracts a flame candidate region based on color information in the target image. Specifically, the flame candidate region extraction unit 101 binarizes pixels in the target image according to whether or not the pixel value indicates a flame-like color. For example, when color information is expressed in RGB (Red, Green, Blue), a flame-like color is a color in which the R value, which is the red component in the color information, is greater than a threshold, or a color in which the difference (RG) value between the red and green components is greater than a threshold.When color information is expressed in HSV (Hue, Saturation, Value, Brightness), a flame-like color is a color in which the H value, which is the hue component, is reddish and the V value, which is the brightness component, is greater than a threshold. The flame candidate region extraction unit 101 extracts a region consisting of a pixel group that exhibits a flame-like color from the binarized pixels, and designates the extracted region as a flame candidate region. In this case, the flame candidate region extraction unit 101 may exclude, as noise, regions having an area less than a threshold from the region consisting of a pixel group that exhibits a flame-like color, and designate regions having an area equal to or greater than the threshold as flame candidate regions. The flame candidate region extraction unit 101 outputs information indicating the extracted flame candidate region to the flame determination unit 102.

[0016] The flame determination unit 102 determines whether a flame is shown in the flame candidate region. The flame determination unit 102 determines whether a flame is shown in the flame candidate region based on an index indicating the complexity of the shape of the flame candidate region. The flame determination unit 102 calculates the index indicating the complexity of the shape of the flame candidate region using statistics of the angle formed by a contour line connecting groups of pixels that make up the contour of the flame candidate region.

[0017] When the shape of a flame candidate region is a simple shape, such as a circle, triangle, rectangle, or polygon, the angles formed by the contour line are concentrated at a specific angle. For example, in the case of a circle, the angle formed by the contour line is an angle depending on the curvature of the circle and the resolution of the image. In the case of a triangle, the angle is either the angle formed by three vertices or the angle formed by a straight line (180 degrees). In the case of a rectangle, the angle is either the angle formed by four vertices (90 degrees for a square or rectangle) or the angle formed by a straight line (180 degrees). For example, a fluorescent light appears on an image as a combination of approximate parallelograms, and the shape of this approximate parallelogram also appears to include the angles formed by the vertices and the angle formed by a straight line (180 degrees) in most parts.

[0018] Flame shapes are generally complex, rather than simple as described above. The contour of a flame usually consists of various angles and is not concentrated at a specific angle. Taking advantage of this complexity of the flame shape, in this embodiment, the flame determination unit 102 calculates an index indicating the complexity of the shape using statistics of the angles formed by the contour line connecting the pixel groups that make up the contour of the flame candidate region. Then, the flame determination unit 102 determines whether a flame is shown in the flame candidate region based on the calculated index. A specific method by which the flame determination unit 102 calculates the angle formed by the contour line will be described in detail later.

[0019] The determination result output unit 103 outputs the determination result by the flame determination unit 102, i.e., whether or not a flame is shown in the target image. The determination result output unit 103 may be configured to display the determination result, for example, by displaying the determination result on a display (not shown). In this case, if a flame is shown in the target image, the user may be alerted by displaying a message indicating the possibility of a fire or by outputting an alarm sound, for example.

[0020] The image information storage unit 104 stores image information. The image information is image information of a target image. The image information may include information indicating the flame candidate region extracted by the flame candidate region extraction unit 101, the determination result determined by the flame determination unit 102, the index used for the determination, and the like. The image information storage unit 104 is configured by a storage medium, such as a hard disk drive (HDD), flash memory, electrically erasable programmable read-only memory (EEPROM), random access read / write memory (RAM), read-only memory (ROM), or any combination of these storage media.

[0021] Here, a specific method by which the flame determination unit 102 calculates the angle formed by the contour line will be described with reference to Fig. 3 (Figs. 3A to 3C). Fig. 3 is a diagram illustrating the processing performed by the flame determination unit 102 according to the embodiment. Each of Figs. 3 shows a target image TG, and schematically shows pixels corresponding to a flame candidate region R extracted from the target image TG. Each of Figs. 3 also shows pixels P1 to P17 as a group of pixels that make up the contour of the flame candidate region R.

[0022] 3A, the flame determination unit 102 selects one pixel, pixel P11 in this example, from the group of pixels P1 to P17 that constitute the outline of the flame candidate region R, as the starting point B. Then, the flame determination unit 102 selects a pixel in one direction from the starting point B and a pixel in the other direction from the group of pixels that constitute the outline of the flame candidate region R. This figure shows an example in which pixel P14, located to the upper left of the starting point B, is selected as point A in one direction, and pixel P8, located to the upper right of the starting point B, is selected as point C in the other direction.

[0023] 3B, the flame determination unit 102 selects three pixels in a combination different from the combination of pixels selected in FIG. 3A. This figure shows an example in which the flame determination unit 102 selects pixel P16 as starting point E. Then, the flame determination unit 102 selects pixel P13, which is located below starting point E, as point D in one direction, and pixel P3, which is located to the right of starting point E, as point F in the other direction.

[0024] 3C, three pixels are selected in a combination different from the combination of pixels selected in FIGS. 3A and 3B. In this example, the flame determination unit 102 selects pixel P7 as the starting point H. Then, the flame determination unit 102 selects pixel P4, which is located to the upper left of the starting point H, as point G in one direction, and pixel P10, which is located to the upper left of the starting point H, as point K in the other direction.

[0025] In this way, the flame determination unit 102 selects three pixels from the group of pixels that form the outline of the flame candidate region R. In this case, the flame determination unit 102 may select three non-adjacent pixels from the group of pixels that make up the outline of the flame candidate region R. Generally, because pixels are arranged in a grid pattern, the angle formed by a line segment connecting three adjacent pixels will be a specific angle, such as 0 degrees, 45 degrees, 90 degrees, 135 degrees, or 180 degrees, and even if the shape of the flame candidate region is complex, the angles formed by the outline will tend to be concentrated at specific angles.

[0026] For example, the flame determination unit 102 selects pixels that are N pixels (N is an integer of 2 or more) apart from the group of pixels that make up the outline of the flame candidate region R. This makes it possible to quantitatively detect the characteristic that when the shape of the flame candidate region is complex, the angles formed by the outline vary, and when the shape of the candidate region is simple, such as a circle or a square, the angles formed by the outline are concentrated at a specific angle.

[0027] Specifically, the flame determination unit 102 extracts, as contour information, a sequence Pn indicating the coordinates of the pixels that make up the contour. The sequence Pn is, for example, a sequence in which the pixels that make up the contour are arranged in order along a predetermined direction, for example, clockwise or counterclockwise. The flame determination unit 102 selects three pixels that are spaced a predetermined distance apart from the sequence Pn. Specifically, the flame determination unit 102 selects three pixels (Pk-d, Pk, Pk+d) that are spaced a distance d apart from the sequence Pn {P1, P2, ..., Pn}.

[0028] The flame determination unit 102 calculates the angle formed by a line segment connecting each of the three selected pixels along the contour. Specifically, the flame determination unit 102 calculates vectors V1 and V2 from the origin Pk to the other points (Pk-d) and (Pk+d) among the coordinates of each of the three pixels (Pk-d, Pk, Pk+d) selected from the sequence Pn. Vector V1 is a vector directed from pixel (Pk) to pixel (Pk-d). Vector V2 is a vector directed from pixel (Pk) to pixel (Pk+d). For example, as shown in equation (1), the flame determination unit 102 calculates the angle θ formed by the line segment using the dot product of the vectors.

[0029] dx1*dx2+dy1*dy2 =√(dx1^2+dy1^2)×√(dx2^2+dy2^2)×cosθ …(1) however, (dx1, dy1): Coordinate values ​​of vector V1 (dx2, dy2): Coordinate values ​​of vector V2 θ: angle between lines

[0030] In this way, the flame determination unit 102 calculates, as the angle of the contour line, the angle formed by the line segment formed along the contour by connecting three pixels selected from the group of pixels that make up the contour of the flame candidate region R. Specifically, the flame determination unit 102 calculates, as the angle formed by the contour line, the angle ABC formed by the line segment ABC in Figure 3A, the angle DFE formed by the line segment DEF in Figure 3B, and the angle GHK formed by the line segment GHK in Figure 3C.

[0031] The flame determination unit 102 calculates the angle of the contour line by, for example, the angle formed by each line segment starting from each of all pixels that make up the contour line.The flame determination unit 102 calculates an index indicating the complexity of the shape using statistics of the angles formed by the contour line.

[0032] The statistical quantity here may be determined arbitrarily as long as it is an index showing whether the angles formed by the contour lines are concentrated at a specific angle, for example, 180 degrees, etc. For example, the statistical quantity may be an average value, a representative value, a mode value, a maximum value, a minimum value, a variance, a standard deviation, or a combination thereof.

[0033] Here, a specific method for calculating the index using statistics by the flame determination unit 102 will be described. In this embodiment, two indexes are used to indicate the complexity of the flame candidate region shape: (1) the concentration level, and (2) the variation in the angle that becomes the most frequent value.

[0034] (1) Concentration is an index that indicates the degree to which the angles of a contour line are concentrated at a specific angle, and is the ratio of the number of angles that form the most frequent value to the total number of angles of the contour line. For example, if line segments are generated starting from each of the pixels that make up the contour and the angles formed by each of the generated line segments are obtained, the total number of angles of the contour line will be the same as the total number of pixels that make up the contour line.

[0035] For example, if the total number of angles made by the contour line is 100, the most common angle is 180 degrees, and there are 67 angles with an angle of 180 degrees, the concentration level is 0.67 (= 67 / 100).Also, if the total number of angles made by the contour line is 100, the most common angle is 7 degrees, and there are 6 angles with an angle of 7 degrees, the concentration level is 0.06 (= 6 / 100).

[0036] (2) The variation in the angle that becomes the most frequent value is an index that indicates the degree to which the most frequent angle formed by the contour line in each image is not concentrated, that is, the degree of variation, when using video images of the monitoring target T. For example, the flame determination unit 102 calculates the most frequent value for each target image that constitutes the video image. The most frequent value here is the angle that becomes the most frequent value among the angles formed by the contour line in the flame candidate region extracted from the target image.

[0037] For example, if a video has 100 target images in total and the most frequent angles in each image are various, such as 17 degrees, 3 degrees, 69 degrees, 41 degrees, etc., the variation will be large. On the other hand, if the most frequent angles in each image are concentrated around 180 degrees, such as 180 degrees, 180 degrees, 181 degrees, 179 degrees, etc., the variation will be small. A statistical quantity such as variance or standard deviation may be used to indicate the degree of variation.

[0038] FIG. 4 is a flowchart showing the flow of processing performed by the flame detection device 10 according to the embodiment. The flame detection device 10 acquires a target image (step S10). The flame detection device 10 acquires, for example, an image of the monitoring target T captured by the camera CA as the target image. Next, the flame detection device 10 extracts a flame candidate region from the target image (step S11). The flame detection device 10 extracts a surface area that shows a flame-like color as a flame candidate region based on color information in the target image. Next, the flame detection device 10 determines whether or not a flame is shown in the flame candidate region extracted in step S11 (step S12). The flame detection device 10 determines whether or not a flame is shown in the flame candidate region based on an index indicating the complexity of the shape of the flame candidate region. Next, the flame detection device 10 outputs the determination result determined in step S12 (step S13). The flame detection device 10 visually displays the determination result and / or outputs an alarm sound according to the determination result.

[0039] FIG. 5 is a flowchart showing step S11 in FIG. 4, that is, the flow of the process of extracting a flame candidate region from a target image. Flame detection device 10 binarizes all pixels in the target image according to whether they have a flame-like color or not (step S110). For example, if the color information is RGB, flame detection device 10 determines that a pixel whose R value is greater than a threshold or whose (RG) value is greater than a threshold is a pixel with a flame-like color. Alternatively, if the color information is expressed in HSV, flame detection device 10 determines that a pixel whose H value is redder and whose V value is greater than a threshold is a pixel with a flame-like color. Flame detection device 10 extracts a region consisting of a group of pixels that exhibit a flame-like color from among the binarized pixels (step S111). Flame detection device 10 selects, from among the extracted regions, a region having an area equal to or greater than a threshold value as a flame candidate region (step S112).

[0040] FIG. 6 is a flowchart showing the process of step S12 in FIG. 4, that is, the process of determining whether or not a flame is shown in the flame candidate region. Flame detection device 10 acquires a group of pixels that form the contour in the flame candidate region extracted in step S11 (step S120). Flame detection device 10 selects three pixels from the acquired group of pixels that form the contour (step S121). Flame detection device 10 calculates the angle formed by a line segment connecting the three pixels along the contour (step S122).

[0041] Flame detection device 10 calculates the angles formed by line segments starting from each of the pixels that make up the contour (step S123). Flame detection device 10 creates, for example, a histogram showing the distribution of angles as statistics of the calculated angles (step S124). Flame detection device 10 creates a histogram that associates angles with the number of corners that fall at each angle, for example, in 5-degree increments. Flame detection device 10 calculates the degree of concentration as an index showing the complexity of the shape based on the created histogram (step S125).

[0042] Flame detection device 10 determines whether the degree of concentration has been calculated for all target images (step S126). If there are target images for which the degree of concentration has not been calculated, flame detection device 10 returns to the process shown in step S10.

[0043] On the other hand, if there is no target image for which the concentration degree has not been calculated, the flame detection device 10 determines whether the concentration degree is less than the threshold value (step S127). For example, the flame detection device 10 determines whether the concentration calculated from each of all target images for which the index has been calculated is less than a threshold value. In this case, for example, if the concentration is less than the threshold value, the flame detection device 10 determines that a flame is shown in the flame candidate region in the target image. Alternatively, the flame detection device 10 may determine whether or not the average value of the degree of concentration calculated from each of all target images for which the index has been calculated is less than a threshold value. If the concentration degree (or the average concentration degree) is less than the threshold, the flame determination unit 102 determines that a flame is shown in the flame candidate region (step S128). On the other hand, if the concentration degree (or the average concentration degree) is equal to or greater than the threshold, the flame determination unit 102 determines that a flame is not shown in the flame candidate region (step S129).

[0044] Although the above flowchart has been described using the degree of concentration as an index, this is not limiting. The degree of variability of the modes calculated from each of the multiple target images constituting the video may also be used as an index. For example, if the variance of the modes calculated from each of the target images is equal to or greater than a threshold, the variability of the modes of each of the target images is large, and thus flame detection device 10 determines that a flame is shown in the flame candidate region. On the other hand, if the variance of the modes calculated from each of the target images is less than the threshold, the variability of the modes of each of the target images is small, and thus flame detection device 10 determines that a flame is not shown in the flame candidate region.

[0045] Here, examples of the target image, flame candidate region, statistics, and index of the embodiment will be described with reference to Figures 7 to 10. Figures 7 to 10 are diagrams for explaining examples of the target image, flame candidate region, statistics, and index of the embodiment. Figures 7 to 10 show the target image TG, a flame candidate region HK extracted from the target image TG, a histogram HG of the angles formed by the contours of the flame candidate region HK, and an index SH (concentration degree) calculated from the statistics of the angles formed by the contours.

[0046] FIG. 7 shows an example of a target image TG in which a flame is captured. The flame candidate region HK extracted from such a target image TG has a complex shape. In this case, the histogram HG of the angles formed by the contour of the flame candidate region HK shows that the flame candidate region HK is composed of angles with various angles ranging from 0 to 180 degrees. In addition, the concentration index is 0.082, and it can be seen from this index that the flame candidate region HK is not concentrated at a specific angle, but has a complex shape.

[0047] FIG. 8 shows an example in which an obscuring object is superimposed on the target image TG shown in FIG. 7. When an obscured object is present, the flame candidate region HK extracted from the target image TG has a shape that does not include the obscured portion. Even in such a case, the histogram HG of the angles formed by the outline of the flame candidate region HK shows that the flame candidate region HK is composed of angles with various angles ranging from 0 to 180 degrees. Furthermore, the concentration index is 0.114, and it can be seen from this index that even if part of the flame candidate region HK is obscured, it can still be determined to be a flame.

[0048] FIG. 9 shows an example of a target image TG in which a circular object is captured. A simple circle is extracted from this target image TG. In this case, the histogram HG of the angles formed by the outline of the flame candidate region HK indicates that there are many specific angles, indicating that the flame candidate region HK is composed of corners with specific angles. The concentration index is 0.516, which indicates that the flame candidate region HK has a simple shape concentrated at a specific angle.

[0049] FIG. 10 shows an example of a target image TG in which a rectangular object is captured. A simple rectangle is extracted from this target image TG. In this case, the histogram HG of the angles formed by the outline of the flame candidate region HK is dominated by a specific angle, indicating that the flame candidate region HK is made up of corners with a specific angle. The concentration index is 0.709, which indicates that the flame candidate region HK has a simple shape concentrated at a specific angle.

[0050] Here, examples of indices when the target image is a moving image will be described with reference to FIG. FIG. 11 shows an example of an index calculated from a video of a flame on the left side, and an example of an index calculated from a video of a car on the right side. The upper part of FIG. 11 shows an example in which the degree of concentration is used as an index, with the horizontal axis representing the frame image number as the time axis and the vertical axis representing the degree of concentration. The bottom of Figure 11 shows an example in which the mode (the most frequent supplementary angle) is used as an index, with the horizontal axis representing the frame image number as the time axis and the vertical axis representing the mode. The supplementary angle here is (180 degrees - angle).

[0051] As shown in the upper part of Figure 11, when a video of a flame is used, the concentration index is less than 0.2 for every image. In contrast, when a video of a car is used, the concentration index is 0.2 or greater for every image. In this way, even for videos captured continuously in chronological order, it is possible to accurately determine whether or not there is a flame using the concentration index.

[0052] The bottom of Figure 11 shows the variation in angle over time using the supplementary angle (= 180 degrees - angle). Since "supplementary angle = 180 degrees - angle," variation in the supplementary angle means variation in the angle. As shown in the lower part of Figure 11, when a video of a flame is used, the mode (the most frequent supplementary angle) varies from image to image, that is, there is a large variation. In contrast, when a video of a car is used, the mode is 0 (zero) for every image, indicating no variation. In this way, even for videos captured continuously in chronological order, it is possible to accurately determine whether or not there is a flame by using the degree of variation in the mode.

[0053] As described above, the flame detection device 10 of the embodiment includes an image acquisition unit 100, a flame candidate region extraction unit 101, and a flame determination unit 102. The image acquisition unit 100 acquires a target image. The flame candidate region extraction unit 101 extracts a flame candidate region from the target image. The flame determination unit 102 determines whether a flame is shown in the flame candidate region. The flame determination unit 102 determines whether a flame is shown in the flame candidate region based on an index indicating the complexity of the shape of the flame candidate region. The flame determination unit 102 calculates the index using statistics of the angle formed by a contour line connecting groups of pixels that make up the contour of the flame candidate region.

[0054] As a result, the flame detection device 10 of this embodiment determines whether a flame is present in a flame candidate region based on an index indicating the complexity of the shape. Because the flame detection device 10 calculates the index using statistics of the angle formed by the contour lines of the flame candidate region, it is possible to prevent regions with simple shapes such as a rectangle from being mistakenly determined to be a flame. Therefore, it is possible to accurately detect flames using a method different from determination based on circularity.

[0055] Furthermore, in the flame detection device 10 of this embodiment, the flame determination unit 102 determines the angle formed by a line segment connecting three pixels selected from a group of pixels constituting the contour of the flame candidate region along the contour as the angle formed by the contour line. The flame determination unit 102 calculates an index using statistics of the angles formed by each of multiple line segments identified for each combination of selected pixels. This allows the flame detection device 10 of this embodiment to determine the angle formed by the contour line as the angle formed by a line segment connecting three pixels selected from a group of pixels constituting the contour of the flame candidate region along the contour, easily calculate the statistics of the angle formed by the contour line, and easily determine whether a flame is indicated in the flame candidate region.

[0056] Furthermore, in the flame detection device 10 of this embodiment, the flame determination unit 102 determines the angle formed by the contour line to be the angle formed by the line segment connecting three non-adjacent pixels selected from the pixel group that forms the contour of the flame candidate region along the contour. This allows the flame detection device 10 of this embodiment to avoid combinations of three adjacent pixels, and to avoid situations in which the angles formed by the contour line appear to be concentrated at a specific angle despite the complex shape, leading to incorrect determination.

[0057] Furthermore, in the flame detection device 10 of this embodiment, the flame determination unit 102 calculates the angle formed by connecting three pixels that are N pixels apart (N is an integer of 2 or more) from the pixel group that makes up the contour, along the contour, as the angle formed by the contour line. This allows the flame detection device 10 of this embodiment to calculate the angle formed by a line segment generated by a combination of three pixels that are spaced at approximately the same intervals, and makes it possible to calculate an index that more accurately indicates the complexity of the shape compared to when the pixel intervals vary, such as being large or small.

[0058] Furthermore, in the flame detection device 10 of this embodiment, the flame determination unit 102 calculates an index as a statistical quantity, which is the concentration level, i.e., the ratio of the number of angles that are the most frequent value to the total number of angles that make up the contour line. As a result, the flame detection device 10 of this embodiment can quantitatively indicate the complexity of the shape as the degree to which the angles that make up the shape are concentrated at a specific angle.

[0059] Furthermore, in the flame detection device 10 of this embodiment, the image acquisition unit 100 acquires moving images, i.e., multiple target images captured in time series of the monitored target. The flame determination unit 102 calculates, as an index, the variation in the most frequent angle among the angles formed by the contour lines corresponding to each of the multiple target images. This allows the flame detection device 10 of this embodiment to determine whether an object has a complex shape that changes from moment to moment as a flame, thereby enabling more accurate determination.

[0060] Furthermore, in the flame detection device 10 of the embodiment, the flame candidate region extraction unit 101 extracts a flame candidate region based on color information in the image. This allows the flame detection device 10 of the embodiment to extract a flame candidate region based on color, making it possible to extract a flame candidate region in a simple manner.

[0061] A modified example will now be described. This modified example differs from the above-described embodiment in that the target image is enlarged or reduced (resized).

[0062] Depending on the distance between the camera CA capturing the target image and the area where the flame has occurred, the outline of the flame may become blurred or the shape of the flame may become too large compared to the size of the image. In this modified example, in such cases, the target image is enlarged or reduced to prevent a decrease in the accuracy of extracting the flame candidate area.

[0063] For example, when installing camera CA, a detection frame of a certain size (say, 10m wide and 5m high) is set in the target image. For example, a first detection frame is set as a frame indicating an area close to camera CA, with pixel coordinates (x, y) corresponding to the four corners being (30, 100), (530, 100), (530, 350), and (30, 350). A second detection frame is set as a frame indicating an area far from camera CA, with pixel coordinates (x, y) corresponding to the four corners being (130, 80), (430, 80), (430, 230), and (130, 230). The direction from the top to the bottom of the image is the positive direction of the y-axis, and the axis perpendicular to the y-axis is the x-axis.

[0064] When such detection frames are set, for example, the size of the first detection frame in the image is 500 x 250 pixels, and the size of the second detection frame is 300 x 150 pixels. When a target image with such detection frames set is acquired, the flame detection device 10 extracts each detection region and enlarges or reduces each extracted detection region to a uniform size, for example, 200 x 125 pixels. This makes the size of the real space captured in one pixel the same regardless of the distance from the camera CA. This prevents the detection results from differing depending on whether a flame of the same size occurs near or far from the camera CA.

[0065] However, when enlarging or reducing an image in a detection frame, if the enlargement / reduction rate is too large, the contours may become unclear. For this reason, it is preferable to set an appropriate combination of the size of each detection frame and the enlargement / reduction rate.

[0066] Flame detection device 10 performs the resizing process described above before extracting a flame candidate region or before determining whether a flame is shown in the flame candidate region. For example, by resizing before extracting a flame candidate region, flame detection device 10 can extract the flame candidate region with high accuracy. Furthermore, by resizing before determining whether a flame is shown in the flame candidate region, it becomes possible to determine with high accuracy whether a flame is shown in the flame candidate region.

[0067] As described above, in the flame detection device 10 according to the modified embodiment, the image acquisition unit 100 acquires a target image having multiple detection frames set according to a combination of the distance from the camera CA and the size of the area in real space. The flame candidate region extraction unit 101 extracts an image region that is within a detection frame from the target image, and enlarges or reduces the extracted image region by a magnification factor that is set in association with the detection frame. The flame candidate region extraction unit 101 extracts a flame candidate region from the image using the image that has been enlarged or reduced by the magnification factor. Furthermore, when the flame candidate region to be judged is a region within the detection frame, the flame judgment unit 102 enlarges or reduces the flame candidate region by a magnification set in association with the detection frame. Using the flame candidate region enlarged or reduced by the magnification, the flame judgment unit 102 judges whether or not a flame is shown in the flame candidate region.

[0068] All or part of the flame detection device 10 in the above-described embodiment may be implemented by a computer. In this case, a program for implementing the functions may be recorded on a computer-readable recording medium, and the program may be loaded and executed by a computer system. Note that the term "computer system" as used herein includes hardware such as an OS and peripheral devices. Furthermore, the term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD-ROMs, as well as storage devices such as hard disks built into a computer system. Furthermore, the term "computer-readable recording medium" may also include devices that dynamically store programs for a short period of time, such as communication lines used when transmitting programs via networks such as the Internet or telephone lines, or devices that store programs for a fixed period of time, such as volatile memory within a computer system serving as a server or client. The program may be for implementing some of the functions described above, or may be capable of implementing the functions in combination with a program already stored in the computer system, or may be implemented using a programmable logic device such as an FPGA.

[0069] Although an embodiment of the present invention has been described above in detail with reference to the drawings, the specific configuration is not limited to this embodiment, and includes designs within the scope of the gist of the present invention. [Explanation of symbols]

[0070] S...Flame detection system 10...Flame detection device 100...Image acquisition unit (acquisition unit) 101...Flame candidate region extraction section (extraction section) 102...Flame judgment section (judgment section)

Claims

1. an acquisition unit that acquires a target image of a monitoring target; an extraction unit that extracts a flame candidate region from the target image acquired by the acquisition unit; a determination unit that calculates an index indicating the complexity of the shape of the flame candidate region using statistics of the angle formed by a contour line connecting groups of pixels that form the contour of the flame candidate region, and determines whether a flame is shown in the flame candidate region based on the calculated index; A flame detection device comprising:

2. the determination unit defines an angle formed by a line segment connecting three pixels selected from a pixel group constituting the contour of the flame candidate region along the contour as the angle formed by the contour line, and calculates the index using statistics of angles formed by each of a plurality of the line segments identified for each combination of the selected pixels.

10. The flame detection device of claim 1.

3. the determination unit determines the angle formed by a line segment connecting three non-adjacent pixels selected from a pixel group constituting the contour of the flame candidate region along the contour as the angle formed by the contour line; 3. The flame detection device of claim 2.

4. the determination unit calculates an angle formed by connecting three pixels that are N pixels apart (N is an integer of 2 or more) from among the pixel group that constitutes the contour along the contour, as the angle formed by the contour line; 4. The flame detection device of claim 3.

5. the determination unit calculates, as the index, a ratio of the number of angles that is the most frequent value to the total number of angles formed by the contour line; A flame detection device according to any one of claims 1 to 4.

6. the acquisition unit acquires a plurality of target images in which the monitoring target is captured in time series, the determining unit calculates, as the index, a variation in an angle that is a most frequent value among angles formed by the contour lines corresponding to each of the plurality of target images; A flame detection device according to any one of claims 1 to 4.

7. the extraction unit extracts the flame candidate region based on color information in the image; A flame detection device according to any one of claims 1 to 6.

8. A flame detection method performed by a flame detection device that is a computer, comprising: The acquisition unit acquires a target image in which the monitoring target is captured, an extraction unit extracting a flame candidate region from the target image acquired by the acquisition unit; a determination unit that calculates an index indicating the complexity of the shape of the flame candidate region using statistics of angles formed by a contour line connecting groups of pixels that form the contour of the flame candidate region, and determines whether a flame is shown in the flame candidate region based on the calculated index; Flame detection methods.

9. The flame detection device is a computer. Acquire an image of the monitored object; extracting a flame candidate region from the acquired target image; calculating an index indicating the complexity of the shape of the flame candidate region using statistics of the angle formed by a contour line connecting groups of pixels that form the contour of the flame candidate region, and determining whether or not a flame is shown in the flame candidate region based on the calculated index; program.

Images