Long-distance fire point small target detection method

Abstract

The invention discloses a long-distance fire point small target detection method, which is characterized in that a plurality of infrared cameras are arranged in a scene, so that the shooting view fully covers the scene, and images are acquired in real time. The method comprises the following steps: processing an image, and judging whether a suspicious fire point small target S exists in a scene or not according to local intensity and local gradient characteristics of a target and a background; and if yes, carrying out continuous n-frame image acquisition on the target S, and calculating the relative change distance of the target S in the continuous n images according to the center of the target S in the n-frame images. And finally, calculating the maximum gray value and the average gray value in an M * M area with the target S center as the center in the n frames of images, and calculating the variance of the M * M area according to the average gray value, so as to judge whether the suspicious fire point small target is the fire point small target. The fire point small target can be found in time, the accuracy is high, and the false alarm rate is low.

Description

Technical Field

[0001] The present invention belongs to the field of infrared imaging processing technology, specifically a long-distance fire point small target detection method, based on the radiation characteristics of the infrared imaging system, is particularly used in forest fire prevention scenes to detect whether there is a fire in the observation field of view, and has the ability to detect small target fire points. Background Art

[0002] Infrared thermal imaging has the advantages of strong penetration and high clarity at night. It can not only penetrate smoke, but also form images in environments with poor visibility or light, or even at night. It can achieve 24-hour uninterrupted monitoring and has been widely used in security, perimeter, military, border and coastal defense and other fields.

[0003] At present, due to the frequent occurrence of forest fires, not only the ecological environment is damaged, but also serious property losses may be caused to nearby residents. Due to the particularity of the forest environment, it is difficult to detect the initial small fire before a large fire occurs, and even forest rangers find it difficult to get close to the depths of the forest. Infrared cameras can not only work 24 hours a day to detect fires, but also have a long range and are less affected by interference. They are more suitable for smoke and fire detection in forest fire prevention, especially when early fires occur, or even before the fire occurs, to provide fire warnings. Although infrared cameras have been widely used in forest fire prevention, there are also cases of false detection and missed detection, especially at long distances, small target fire targets are easily missed, and the opportunity for rescue is missed, resulting in immeasurable losses. Based on infrared imaging technology, this patent proposes a detection method for small targets of long-distance fire points for forest fire prevention. This method can not only accurately detect the target, but also reduce the probability of false detection.

[0004] In general, infrared small target detection is a challenging task for computer vision. In infrared images, infrared small targets are in the shape of isotropic Gaussian intensity functions. In terms of intensity, the brightness value of the target is greater than the values ​​of its neighboring pixels in the infrared image. For a two-dimensional Gaussian function, almost all gradients of the function point to its center. Similarly, the corresponding gradients of infrared small targets also roughly point to the center of the target. These two characteristics are called local intensity characteristics and local gradient characteristics, respectively. In infrared images, since the grayscale values ​​of the background are almost close, their intensity values ​​are also close, and the intensity value of the target is greater than that of the background, so the uniform background can be effectively suppressed by the local intensity threshold; similarly, for backgrounds with strong edges, their gradient directions are usually consistent, and these gradients are significantly different from the gradients of the target in distribution. Therefore, by combining these two characteristics, local intensity characteristics and local gradient characteristics, background clutter can be effectively suppressed and high-temperature targets can be extracted. Summary of the invention

[0005] In view of the problems existing in the detection of small fire targets in existing forest fire prevention scenarios, the present invention proposes a long-distance small fire target detection method, which can detect whether there is a fire in the forest as quickly as possible, so as to facilitate timely response measures.

[0006] The method for detecting small targets at long-distance fire points of the present invention comprises the following specific steps:

[0007] Step 1: Place infrared cameras in the scene to be monitored and collect data in real time.

[0008] Step 2: Process the acquired image to determine the target and background areas, and further calculate the local intensity characteristics Q of the target and background areas. p and the local gradient characteristic T p ; Determine whether there is a small target that can be ignited in the image.

[0009] Step 3: Determine whether there is a small fire target in the image.

[0010] When the local intensity characteristic Q p Greater than a given threshold Q Y , and the local gradient characteristic T p The gradient directions all point to the center of a target, then the local gradient feature T p The target pointed by the gradient direction is marked as S, which is a small target of a suspicious fire point. At the same time, the current moment is recorded as the moment t when the target is discovered.

[0011] Step 4: After the small fire targets are found in step 3, the infrared camera focuses on each small fire target and continuously collects n frames of infrared images, which are marked as N1, N2, ..., N respectively. n .

[0012] Step 5: Perform the above steps 2 and 3 on the collected n frames of infrared images to obtain N small targets S marked with suspicious fire points. n Frame image.

[0013] Step 6: Record N1, N2, ..., N n The center coordinates of the small target S of the suspicious fire point in the frame image are Z1, Z2, ..., Z n .

[0014] Step 7: Calculate the coordinate change L of the center position of the suspicious small target fire point S in the continuous n frames of images 21 =|Z2-Z1|, L 31 =|Z3-Z1|, ..., L n1 =|Z n -Z1|.

[0015] Step 8: Calculate the maximum grayscale value of the suspicious small target S in the M*M area centered on the center coordinate Z1 in the n consecutive images, marked as M1, M2, ..., M n ; Calculate the grayscale mean of the M*M size area, marked as J1, J2, ..., J n According to the mean values ​​J1, J2, …, J n Calculate the variance of the M*M size area of ​​multiple images, marked as V.

[0016] Step 9: When the distance L changes 21 , L 31 …、L n1 are all less than the given threshold L Y , maximum values ​​M1, m2, ..., M n are all greater than the maximum given threshold M Y , the variance is V greater than the given variance threshold V Y At this time, the suspicious small target S is considered to be a small fire point target, and a fire warning is issued to remind the staff to check whether there is a fire on the scene.

[0017] The advantages of the present invention are:

[0018] 1) The long-distance fire point small target detection method of the present invention utilizes the advantages of the infrared camera's long working distance and 24-hour uninterrupted operation to detect small fire point targets in forest fire prevention scenes. It can not only intuitively display the real-time infrared image of the forest scene, but also detect small fire point targets in time with high accuracy and low false alarm rate.

[0019] 2) Compared with the traditional AI-based fire identification solution, the long-distance fire point small target detection method of the present invention has a long working distance and can identify small target fire points smaller than 5*5 pixels. It does not require AI calculation or visible light camera, but only infrared camera, which greatly reduces the equipment cost.

[0020] 3) The long-distance fire point small target detection method of the present invention can timely discover the long-distance fire point small target, and then take timely measures to reduce the loss to a minimum. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is a flow chart of the long-distance fire point small target detection method of the present invention. DETAILED DESCRIPTION

[0022] The present invention will be further described in detail below in conjunction with the accompanying drawings.

[0023] The long-distance fire point small target detection method of the present invention is based on the design of infrared thermal imaging system and can identify small target fire points smaller than 5*5 pixels in forest fire prevention areas, such as Figure 1 As shown, the specific steps are:

[0024] Step 1: Camera Arrangement

[0025] Infrared thermal imaging cameras, based on the customer's requirements for scene image clarity and cost control, choose infrared thermal imaging cameras with different resolutions. For example, if the clarity requirement is high and the cost allows, choose a resolution of 1920*1280; if the clarity requirement is not high and the cost is limited, choose a resolution of 384*288; if the clarity requirement is moderate and the cost is moderate, choose a resolution of 640*512.

[0026] The infrared thermal imaging camera can be selected as cooled or uncooled type, has temperature measurement function, real-time gray value analysis function, and real-time upload danger signal alarm function.

[0027] The infrared thermal imaging camera is fixed on a bracket. According to the field of view of the forest scene to be observed, N infrared cameras are set in the scene to ensure that the field of view of the N cameras covers the entire scene area.

[0028] Step 2: Obtain target and background features in the image

[0029] N infrared thermal imaging cameras acquire scene image data within their shooting range in real time, and compare the image grayscale value with the grayscale value threshold set by experience. The image with a grayscale greater than the threshold is the target area, and the image with a grayscale less than the threshold is the background area. Furthermore, the local intensity characteristics Q of the target area and the background area in each image are calculated. p and the local gradient characteristic T p .

[0030] Step 3: Based on the local strength characteristic Q obtained in step 2 p and the local gradient characteristic T p Determine whether there are small targets that can be ignited in the image.

[0031] When the local intensity characteristic Q p Greater than a given threshold Q Y , and the local gradient characteristic T p The gradient directions all point to the center of a target, then the local gradient feature T p The target pointed by the gradient direction is marked as S, which is a small target of a suspicious fire point. At the same time, the current moment is recorded as the moment t when the target is discovered.

[0032] Step 4: After the small fire targets are found in step 3, the infrared camera focuses on each small fire target and continuously collects 25 to 30 frames of infrared images, which are marked as N1, N2, ..., N respectively. n .

[0033] Step 5: Perform the above steps 2 and 3 on the collected n frames of infrared images to obtain N small targets S marked with suspicious fire points.n Frame image.

[0034] Step 6. Record N1, N2, ..., N n The center coordinates of the small target S of the suspicious fire point in the frame image are Z1, Z2, ..., Z n .

[0035] Step 7: Calculate the coordinate change L of the center position of the suspicious small target fire point S in the continuous n frames of images 21 =|Z2-Z1|, L 31 =|Z3-Z1|, ..., L n1 =|Z n -Z1|.

[0036] Step 8: Calculate the maximum grayscale value of the suspicious small target S in the M*M area centered on the center coordinate Z1 in the n consecutive images, marked as M1, M2, ..., M n ; M is 80, which is determined by the resolution. Calculate the grayscale mean of the M*M size area, marked as J1, J2, ..., J n According to the mean values ​​J1, J2, …, J n Calculate the variance of the M*M size area of ​​multiple images, marked as V.

[0037] Step 9: When the distance L changes 21 , L 31 …、L n1 are all less than the given threshold L Y , the maximum values ​​of M1, M2, ..., M n are all greater than the maximum given threshold M Y , the variance is V greater than the given variance threshold V Y At this time, the suspicious small target S is considered to be a small fire point target, and the system issues a fire warning to remind the staff to check whether there is a fire on site.

Claims

1. A method for detecting small targets at long distance fire points, characterized in that: The specific steps are: Step 1: Arrange infrared cameras in the scene to be monitored and collect data in real time; Step 2: Process the acquired image to determine the target and background areas, and further calculate the local intensity characteristics Q of the target and background areas. p and the local gradient characteristic T p ; Determine whether there is a small target that can be ignited in the image; Step 3: Determine whether there is a small fire target in the image; When the local intensity characteristic Q p Greater than a given threshold Q Y , and the local gradient characteristic T p The gradient directions all point to the center of a target, then the local gradient feature T p The target pointed by the gradient direction is marked as S, which is a small target of a suspected fire point. At the same time, the current moment is recorded as the moment t at which the target is found. Step 4: After the small fire targets are found in step 3, the infrared camera focuses on each small fire target and continuously collects n frames of infrared images, which are marked as N1, N2, ..., N respectively. n ; Step 5: Perform the above steps 2 and 3 on the collected n frames of infrared images to obtain N small targets S marked with suspicious fire points. n Frame image; Step 6. Record N1, N2, ..., N n The center coordinates of the small target S of the suspicious fire point in the frame image are Z1, Z2, ..., Z n ; Step 7: Calculate the coordinate change L of the center position of the suspicious small target fire point S in the continuous n frames of images 21 =|Z2-Z1|, L 31 =|Z3-Z1|, ..., L n1 =|Z n -Z1|; Step 8: Calculate the maximum grayscale value of the suspicious small target S in the M*M area centered on the center coordinate Z1 in the n consecutive images, marked as M1, M2, ..., M n ; Calculate the grayscale mean of the M*M size area, marked as J1, J2, ..., J n . According to the mean values ​​J1, J2, ..., J n Calculate the variance of the M*M size area of ​​multiple images, marked as V; Step 9: When the distance L changes 21 , L 31 ...、L n1 are all less than the given threshold L Y , the maximum values ​​of M1, M2, ..., M n are all greater than the maximum given threshold M Y , the variance is V greater than the given variance threshold V Y At this time, the suspicious small target S is considered to be a small fire point target, and a fire warning is issued to remind the staff to check whether there is a fire on the scene.

2. A method for detecting small targets at long distance fire points as claimed in claim 1, characterized in that: The infrared thermal imaging camera can be selected as cooled or uncooled type, has the function of temperature measurement, real-time analysis of grayscale value, and real-time upload of danger signal alarm.

3. A method for detecting small targets at long distance fire points as claimed in claim 1, characterized in that: The infrared thermal imaging camera is fixed on a bracket. According to the field of view of the scene area to be observed, N infrared cameras are set in the scene to ensure that the field of view of the N cameras covers the entire scene area.

4. A method for detecting small targets at long distance fire points as claimed in claim 1, characterized in that: In step 2, the grayscale value of the collected image is compared with the grayscale value threshold. The image with a grayscale greater than the threshold is the target area, otherwise it is the background area.

5. A method for detecting small targets at long distance fire points as claimed in claim 1, characterized in that: In step 4, it is appropriate to collect 25 to 30 frames of images.

Images