Aircraft dry cabin fire detection method fusing temperature field and improved YOLO model

Abstract

The invention relates to the technical field of airplane dry cabin fire detection, in particular to an airplane dry cabin fire detection method fusing a temperature field and an improved YOLO model, and solves the problem that an existing airplane dry cabin fire early-stage detection technology lacks. Processing the thermal imaging image to form a temperature field image of a monitoring area, extracting an overtemperature area exceeding a temperature threshold value, extracting a visible light image of the same area, fusing the visible light image and the thermal imaging image, inputting the fused image into an improved YOLO model to carry out fire hazard judgment, sending out a fire alarm if flames or smog exists, and otherwise, sending out a fire alarm if the flames or smog exists. And continuing to detect the next frame of image. The system is used for early warning and visual warning of the fire disaster of the dry cabin of the airplane.

Description

Technical Field

[0001] The present invention relates to the technical field of aircraft dry compartment fire detection, and in particular to an aircraft dry compartment fire detection method integrating a temperature field and an improved YOLO model. Background Art

[0002] Aircraft dry compartment refers to the compartment or cavity between the outer skin and the internal structure of the aircraft. Hydraulic pipelines, fuel pipelines, avionics equipment and lines are usually distributed in the dry compartment. If a fire occurs in the dry compartment of an aircraft, it is difficult to extinguish the fire due to the confined space, and the danger is relatively large, which is related to the flight safety of the entire aircraft. Therefore, it is necessary to accurately and quickly detect the fire point in the early stage of the dry compartment fire and issue an audible and visual alarm.

[0003] Existing aircraft fire detection devices cannot be directly applied to aircraft dry compartments, and the fire detection function is single and cannot provide a visual alarm function. Summary of the invention

[0004] The purpose of the present invention is to provide an aircraft dry compartment fire detection method integrating temperature field and improved YOLO model, which is used for early warning and visual alarm of aircraft dry compartment fire, solves the problem of lack of existing aircraft dry compartment fire early detection technology, and improves the safety of aircraft during flight.

[0005] The technical solution of the present invention:

[0006] A method for detecting aircraft dry compartment fire by integrating temperature field and improved YOLO model.

[0007] Visible light images and thermal imaging images are obtained through the dry cabin dual-light camera, and the thermal imaging image is processed to form a temperature field image of the monitored area. The over-temperature area exceeding the temperature threshold is extracted, and then the visible light image of the same area is extracted. The visible light image and the thermal imaging image are fused, and the fused image is input into the improved YOLO model for fire judgment. If there is flame or smoke, a fire alarm is issued, otherwise, the next frame of image is detected.

[0008] The thermal imaging image is processed to form a temperature field image of the monitored area, and then the over-temperature area exceeding the temperature threshold is extracted, including:

[0009] (1) The dual-light camera collects the thermal imaging image and adjusts the thermal imaging image to 400×300 pixels;

[0010] (2) Filter out pixels that exceed the temperature threshold. The specific method is as follows:

[0011]

[0012] Where Px represents the pixel of the image;

[0013] (3) Use the smallest rectangular frame to surround the pixels with too high temperature, as shown below:

[0014] (x,y,w,h)=min{}

[0015] In the formula, x and y represent the horizontal and vertical coordinates of the center point of the minimum rectangular box, respectively, and w and h represent the width and height of the minimum rectangular box, respectively;

[0016] The fusion of the visible light image and the thermal imaging image is specifically as follows: superimposing the area corresponding to the rectangular frame of the visible light image and the area corresponding to the rectangular frame of the thermal imaging, marking the area with the highest temperature in the visible light image, and displaying the temperature value at the same time.

[0017] The method for fusing visible light images and thermal imaging images is as follows

[0018]

[0019] In the formula, Represents the pixel point after the fusion of visible light and thermal imaging images, Represents the target pixel of the visible light image, represents the pixel point in the thermal imaging target area, λ and μ represent the weights of the visible light and thermal imaging images respectively.

[0020] The fire alerted the pilots via displays in the cockpit.

[0021] The improved YOLO model adopts a deep learning network model.

[0022] The improved YOLO model adds a GAM Attention module.

[0023] The improved YOLO model is trained by a large number of aircraft dry cabin fire image data sets to obtain parameters of the network model.

[0024] Beneficial effects of the present invention: The present invention realizes early rapid detection and visual alarm of dry cabin fire, and improves the safety of the aircraft during flight. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a schematic diagram of an aircraft dry compartment fire detection method integrating a temperature field and an improved YOLO model according to the present invention.

[0026] Figure 2 It is a schematic diagram of the network model parameter training process of the improved YOLO model of the present invention.

[0027] Figure 3 It is a schematic diagram of the GAM Attention module model structure added in the present invention.

[0028] Figure 4 It is a schematic diagram of the improved YOLO model structure of the present invention.

[0029] Figure 5 This is a flame recognition effect diagram after image fusion of the present invention. DETAILED DESCRIPTION

[0030] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0031] like Figure 1 As shown, the present invention provides an aircraft dry compartment fire detection method that integrates temperature field and improved YOLO model. The specific implementation method is as follows: A method for detecting aircraft dry compartment fire that integrates temperature field and improved YOLO model, obtains visible light image and thermal imaging image through dry compartment dual-light camera, processes the thermal imaging image to form a temperature field image of the monitoring area, and then extracts the over-temperature area exceeding the temperature threshold, then extracts the visible light image of the same area, and then fuses the two types of images, and the fused image is input into the improved YOLO model for fire judgment. If there is flame or smoke, a fire alarm is issued, otherwise, continue to detect the next frame of image.

[0032] Furthermore, the over-temperature region extraction method is:

[0033] (1) The dual-light camera collects the thermal imaging image and adjusts the thermal imaging image to 400×300 pixels;

[0034] (2) Filter out pixels that exceed the temperature threshold. The specific method is as follows:

[0035]

[0036] In the formula, Px represents the pixel point of the image. When a fire occurs, the temperature will gradually rise, and the temperature during a fire is much higher than the surrounding background temperature. Therefore, this module first selects some pixels with the highest temperature in the thermal imaging image.

[0037] (3) Use the smallest rectangular frame to surround the pixels with too high temperature, as shown below:

[0038] (x,y,w,h)=min{}

[0039] Where x and y represent the horizontal and vertical coordinates of the center point of the minimum rectangular box, respectively, and w and h represent the width and height of the minimum rectangular box, respectively.

[0040] Furthermore, the image fusion method is to superimpose the area corresponding to the visible light image rectangle frame with the area corresponding to the thermal imaging rectangle frame, mark the area with the highest temperature in the visible light image, and display the temperature value at the same time. The specific method is as follows

[0041]

[0042] In the formula, Represents the pixel point after the fusion of visible light and thermal imaging images, Represents the target pixel of the visible light image, represents the pixel point in the thermal imaging target area, λ and μ represent the weights of the visible light and thermal imaging images, respectively, which are used to adjust the sensitivity.

[0043] like Figure 2-Figure 4 As shown in the figure, the improved YOLO model further performs fire recognition on the fused image. The module adopts a deep learning network model, which includes an input end, a backbone network, a neck network and an output end. The input layer is used to input the detection image. The main function of the backbone network is to extract features. The main structure includes CBS module, C3 module and SPPF module; CBS module and C3 module are composed of convolution kernels, which are mainly used to extract features and reduce the dimension of feature maps. SPPF module is a pyramid structure mainly used to fuse multi-scale features. The main function of the neck network is to combine shallow image features with deep semantic features to obtain more complete features. The main structure includes C3 module, upsample module and concat module. The upsample module is used to enlarge the size of the feature map and restore the resolution of the feature map. The concat module is used to splice multiple tensors according to the specified dimension. The output layer is mainly responsible for receiving the feature map processed and enhanced by the neck network, and performing the final target classification and bounding box regression operations. At the same time, the GAM Attention module is introduced into the model to improve the recognition accuracy. The GAM module consists of two key components: channel attention module and spatial attention module. Channel attention models the importance of each feature channel, assigns different weights to each channel, suppresses the learning of redundant features, and enhances the feature recognition of flame and smoke targets; in the spatial attention module, two convolutional layers are used for spatial information fusion, which increases the parameter learning of spatial information. The GAM module combines the advantages of the channel attention module and the spatial attention module to improve the accuracy of flame and smoke target recognition. The model is trained with a large number of aircraft dry cabin fire image data sets to obtain the parameters of the network model, which can accurately identify whether there is smoke and flame in the dry cabin monitoring video. If so, Figure 5If shown, an alarm is issued.

[0044] The above is only a specific embodiment of the present invention, and the present invention is described in detail. The unexplained part is a conventional technology. However, the protection scope of the present invention is not limited to this. Any changes or substitutions that can be easily thought of by any technician familiar with the technical field within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. The protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for detecting aircraft dry compartment fire by integrating temperature field and improved YOLO model, characterized in that: Visible light images and thermal imaging images are obtained through the dry cabin dual-light camera, and the thermal imaging image is processed to form a temperature field image of the monitored area. The over-temperature area exceeding the temperature threshold is extracted, and then the visible light image of the same area is extracted. The visible light image and the thermal imaging image are fused, and the fused image is input into the improved YOLO model for fire judgment. If there is flame or smoke, a fire alarm is issued, otherwise, the next frame of image is detected.

2. The aircraft dry compartment fire detection method integrating temperature field and improved YOLO model as claimed in claim 1, characterized in that: The thermal imaging image is processed to form a temperature field image of the monitored area, and then the over-temperature area exceeding the temperature threshold is extracted, including: (1) The dual-light camera collects the thermal imaging image and adjusts the thermal imaging image to 400×300 pixels; (2) Filter out pixels that exceed the temperature threshold. The specific method is as follows: Px=max{temperature} Where Px represents the pixel of the image; (3) Use the smallest rectangular frame to surround the pixels with too high temperature, as shown below: (x,y,w,h)=min{} Where x and y represent the horizontal and vertical coordinates of the center point of the minimum rectangular box, respectively, and w and h represent the width and height of the minimum rectangular box, respectively.

3. The aircraft dry compartment fire detection method integrating temperature field and improved YOLO model as claimed in claim 1, characterized in that: The fusion of the visible light image and the thermal imaging image is specifically as follows: superimposing the area corresponding to the rectangular frame of the visible light image and the area corresponding to the rectangular frame of the thermal imaging, marking the area with the highest temperature in the visible light image, and displaying the temperature value at the same time.

4. The aircraft dry compartment fire detection method integrating temperature field and improved YOLO model as claimed in claim 3, characterized in that: The method for fusing visible light images and thermal imaging images is as follows In the formula, Represents the pixel point after the fusion of visible light and thermal imaging images, Represents the target pixel of the visible light image, represents the pixel point in the thermal imaging target area, λ and μ represent the weights of the visible light and thermal imaging images respectively.

5. The aircraft dry compartment fire detection method integrating temperature field and improved YOLO model as claimed in claim 1, characterized in that: The fire alerted the pilots via displays in the cockpit.

6. The aircraft dry compartment fire detection method integrating temperature field and improved YOLO model as claimed in claim 1, characterized in that: The improved YOLO model adopts a deep learning network model.

7. The aircraft dry compartment fire detection method integrating temperature field and improved YOLO model as claimed in claim 6, characterized in that: The improved YOLO model adds a GAM Attention module.

8. The aircraft dry compartment fire detection method integrating temperature field and improved YOLO model as claimed in claim 7, characterized in that: The improved YOLO model is trained by a large number of aircraft dry cabin fire image data sets to obtain parameters of the network model.

Images