A method and system for intelligent wearable device danger perception based on infrared technology

By combining infrared and color image technology on smart wearable devices, using deep learning and genetic algorithms to identify risk factors in the wild night environment and dynamically adjust the path, the problem of the existing technology being unable to identify and deal with environmental hazards is solved, and the problem of the current technology being unable to identify and deal with environmental hazards is achieved. The safety of walking in the wild at night is improved.

CN117746556BActive Publication Date: 2024-11-22JIANGSU ZHONGKE LANGRUN INTELLIGENT TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311600076.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-11-28
Publication Date
2024-11-22
Estimated Expiration
2043-11-28

AI Technical Summary

Technical Problem

The existing hazard perception methods cannot effectively identify the risk factors in the environment and adjust the user path in the wild night environment, resulting in safety hazards.

Method used

Using intelligent wearable devices based on infrared technology, images are collected through infrared image acquisition devices and color image acquisition cards, using mean method, gradient field fusion weights, multi-scale transformation, deep learning models and genetic algorithms to identify risk factors in real time and adjust their walking dynamically path.

Benefits of technology

Effectively identify potential hazards in dark environments, dynamically recommend the safest paths, timely warnings and re-planning routes, improving the safety of outdoor night walking.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117746556B_ABST
    Figure CN117746556B_ABST
Patent Text Reader

Abstract

The present invention discloses a danger perception method and system for intelligent wearable devices based on infrared technology, which relates to the field of danger alarm technology. The method comprises the following steps: collecting sub-region images and identifying danger factors in the sub-region images; configuring sensors on the intelligent wearable device to collect position data; planning several walking routes based on the activity area, and recommending walking paths to the wearer in real time, and if danger factors appear on the wearer's walking path, an early warning is issued and a walking path is re-recommended; if the wearer enters a sub-region where danger factors appear, an alarm is issued through the intelligent wearable device; the system comprises an image acquisition module, a position data acquisition module, a path early warning module and an alarm module. The safety of users walking at night or in low-light environments is ensured, and the probability of users encountering danger is significantly reduced through early warning and dynamic path planning.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of danger alarm technology, and in particular to a danger perception method and system for intelligent wearable devices based on infrared technology. Background Art

[0002] With the development of science and technology, people are paying more and more attention to personal safety, especially in certain specific environments, such as construction sites, outdoor areas or night travel. Therefore, the hazard perception technology of smart wearable devices has begun to attract attention, among which the method based on infrared technology is particularly important.

[0003] Infrared technology uses the radiation properties of infrared light for detection. Compared with visible light, infrared light has a longer wavelength, usually between 0.7-1000 microns. Infrared light mainly comes from the thermal radiation of objects, so it is particularly effective for detecting temperature changes or hot moving bodies. In the past, people mainly relied on visible light cameras, sound detection or vibration detection for safety detection. But these methods do not work well in dim light or obstructed conditions. For example, at night or in thick fog, the field of view of traditional cameras may be limited, and sound detection may be interfered with by environmental noise. Infrared technology can work in no light or low light conditions because it relies on the thermal radiation emitted by objects rather than reflected light. This means that infrared technology can still be effectively detected in environments such as night or thick fog. Therefore, applying infrared technology to the hazard perception method of smart wearable devices will bring many benefits.

[0004] Among the existing danger perception methods, for example, Chinese patent 202210845661.9 discloses a method for dynamic perception of dangerous situations of power workers that integrates multi-dimensional information. It can perform danger alarms or situation perception warnings based on the recognition results in multiple dimensions, and realizes comprehensive dynamic perception and warning of dangerous situations of power workers, effectively reducing the probability of safety accidents. However, this method still has the following shortcomings: for outdoor night environments, in order to ensure walking safety, the path needs to be planned in advance, while the existing dynamic perception of dangerous situations only identifies the status of personnel, does not process environmental data, and does not use the dangerous factor variables in the environment to warn users and adjust the user's path planning.

[0005] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0006] In view of the problems in the related art, the present invention proposes a hazard perception method and system for a smart wearable device based on infrared technology to overcome the above-mentioned technical problems existing in the existing related technology.

[0007] To this end, the specific technical solution adopted by the present invention is as follows:

[0008] According to one aspect of the present invention, a method for intelligent wearable device danger perception based on infrared technology is provided, the method comprising the following steps:

[0009] S1. Several infrared image acquisition devices and color image acquisition cards are configured in the activity area, and sub-area images are acquired by the infrared image acquisition devices and color image acquisition cards in each sub-area, and risk factors are identified in the sub-area images.

[0010] S2. A smart wearable device is configured on the wearer, and a sensor is configured on the smart wearable device to collect location data.

[0011] S3. Plan several walking routes based on the activity area and recommend walking paths to the wearer in real time. If dangerous factors appear on the wearer's walking path, issue an early warning and recommend a new walking path.

[0012] S4. If the wearer enters a sub-area where dangerous factors occur, an alarm is issued through the smart wearable device.

[0013] Furthermore, the configuration of a plurality of infrared image acquisition devices and color image acquisition cards in the activity area, and the acquisition of sub-area images by the infrared image acquisition devices and color image acquisition cards in each sub-area, and the identification of risk factors in the sub-area images includes the following steps:

[0014] S11, in each sub-area, an infrared thermal image in the sub-area is acquired by an infrared image acquisition device, and a color image in the sub-area is acquired by a color image acquisition card;

[0015] S12. According to the characteristics of the infrared thermal image, the mean value method is used to extract the features of the temperature rise area to obtain a source image feature map;

[0016] S13, using the source image feature map to calculate the gradient field fusion weight, and using the multi-scale transformation of Gaussian filtering combined with the image gradient covariance to obtain the structural features of the infrared thermal image;

[0017] S14, select the scale factor;

[0018] S15, using a soft-max function to soften the gradient weight of the edge area of ​​the infrared thermal image, and obtaining a structure tensor with improved gradient weight;

[0019] S16, obtaining a reconstructed image according to the variational technique and the Poisson equation solution;

[0020] S17, using a transparency method to fuse the color image and the reconstructed image to obtain a sub-region image;

[0021] S18, extracting features of the sub-region image using image processing technology, and establishing historical data based on the features;

[0022] S19. Establish a deep learning model, and use historical data to train the deep learning model, input the features of the new sub-region image into the trained deep learning model, and detect risk factors.

[0023] Furthermore, when the gradient field fusion weight is calculated using the source image feature map, the gradient field fusion weight is:

[0024]

[0025] In the formula, S n Represents the feature map of the nth source image;

[0026] S k Represents the feature map of the k-th source image;

[0027] N represents the total number of source images.

[0028] Further, the selecting the scale factor comprises the following steps:

[0029] S141. Comparative analysis of multi-scale factors on infrared thermal images;

[0030] S142: Select a scale factor according to the richness of image detail information.

[0031] Furthermore, the planning of several walking routes based on the activity area and the recommendation of walking paths to the wearer in real time, and if a risk factor appears on the wearer's walking path, an early warning is issued and a new walking path is recommended, including the following steps:

[0032] S31, taking each sub-area of ​​the activity area as a node of the graph, taking the road between any two nodes as an edge of the graph, and assigning weights to the edges according to the distance between the two points, the estimated travel time, and the congestion level of the road;

[0033] S32, update edge weights in real time;

[0034] S33, according to the current graph weight and based on the genetic algorithm, obtaining the optimal path from the starting point to the end point;

[0035] S34, if the optimal path is determined, the optimal path is sent to the wearer's smart wearable device in real time by wireless transmission, and the wearer is prompted to walk along the optimal path;

[0036] S35, if the wearer deviates from the optimal path, the optimal path is recalculated and updated and pushed to the wearer;

[0037] S36. If a dangerous factor appears on any edge of the wearer's walking path, an early warning is issued, the weight of the edge is reduced, the optimal path is recalculated, and the update is pushed to the wearer.

[0038] Furthermore, the real-time updating of edge weights includes the following steps:

[0039] S321, collecting road information in the activity area in real time, and dividing the road information into positive information and negative information;

[0040] S322. If negative information appears, the weight of the corresponding edge in the graph is increased, and the cost increases as the weight increases.

[0041] Furthermore, the step of obtaining the optimal path from the starting point to the end point based on the current graph weight and the genetic algorithm includes the following steps:

[0042] S331, encoding the walking path into chromosomes, and each chromosome represents a path;

[0043] S332, randomly generate a number of chromosomes and form a primary population;

[0044] S333, calculating the total cost of each path according to the weight of the edge in the graph;

[0045] S334, evaluating each chromosome according to the total cost through the fitness function, and the lower the total cost, the higher the fitness;

[0046] S335, setting a termination condition;

[0047] S336: If the termination condition is triggered, the chromosome with the highest fitness is selected and decoded into a path, and pushed to the wearer's smart wearable device.

[0048] Furthermore, when each chromosome is evaluated according to the total cost through the fitness function, the fitness function is:

[0049]

[0050] In the formula, D represents the total cost of each path;

[0051] ε represents a positive number that is divisible by zero.

[0052] Furthermore, if the wearer enters a sub-area where dangerous factors occur, the alarm is issued by the smart wearable device, including the following steps:

[0053] S41, determining the time length between the time when the risk factor appears and the current time, if the time length is less than a preset threshold, determining that the risk factor is valid;

[0054] S42. If the risk factor is valid, an alarm is issued through the smart wearable device.

[0055] According to another aspect of the present invention, there is provided a danger perception system for intelligent wearable devices based on infrared technology, the system comprising an image acquisition module, a position data acquisition module, a path warning module and an alarm module;

[0056] Wherein, the image acquisition module and the location data acquisition module are both connected to the path warning module and the alarm module.

[0057] The image acquisition module is used to configure a plurality of infrared image acquisition devices and color image acquisition cards in the activity area, and to acquire sub-area images in each sub-area through the infrared image acquisition devices and color image acquisition cards, and to identify risk factors in the sub-area images.

[0058] The position data collection module is used to configure a smart wearable device on the wearer, and configure a sensor on the smart wearable device to collect position data.

[0059] The path warning module is used to plan several walking routes based on the activity area and recommend walking routes to the wearer in real time. If a dangerous factor appears on the wearer's walking path, an early warning is issued and a walking route is re-recommended.

[0060] The alarm module is used to issue an alarm through the smart wearable device if the wearer enters a sub-area where dangerous factors appear.

[0061] The beneficial effects of the present invention are:

[0062] (1) The present invention provides a method and system for intelligent wearable device danger perception based on infrared technology, which solves the safety problem of walking in the wild at night, especially for those possible environmental hazards. The present invention combines infrared technology with color imaging technology, which can not only capture images in the dark, but also identify potential dangers based on temperature differences and infrared characteristics, and fuse the color image with the reconstructed image to obtain a sub-region image, so that each sub-region image makes up for the problem of lack of detailed information in a simple infrared thermal image, which is particularly useful in the wild environment and can identify animals, protruding obstacles or other heat sources.

[0063] (2) The present invention can dynamically recommend the safest and fastest path for users, and timely re-plan the route when danger occurs, ensuring that users always walk on the safest path. By using genetic algorithms, the optimal solution can be found quickly and efficiently among multiple possible paths, which is of great value in a changing environment. When users approach or enter a dangerous area, the smart wearable device will provide timely warnings to help users avoid potential risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0065] Figure 1 is a flow chart of a danger perception method for a smart wearable device based on infrared technology according to an embodiment of the present invention;

[0066] Figure 2 The invention is a block diagram of a danger perception system for a smart wearable device based on infrared technology according to an embodiment of the invention.

[0067] In the figure:

[0068] 1. Image acquisition module; 2. Location data acquisition module; 3. Path warning module; 4. Alarm module. DETAILED DESCRIPTION

[0069] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in the field should be able to understand other possible implementations and advantages of the present invention. The components in the figures are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0070] According to an embodiment of the present invention, a method and system for hazard perception of a smart wearable device based on infrared technology are provided.

[0071] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to one embodiment of the present invention, a method for hazard perception of a smart wearable device based on infrared technology is provided, and the method comprises the following steps:

[0072] S1. Several infrared image acquisition devices and color image acquisition cards are configured in the activity area, and sub-area images are acquired by the infrared image acquisition devices and color image acquisition cards in each sub-area, and risk factors are identified in the sub-area images.

[0073] In a further embodiment, the plurality of infrared image acquisition devices and color image acquisition cards are configured in the activity area, and sub-area images are acquired in each sub-area by the infrared image acquisition devices and color image acquisition cards, and identifying risk factors in the sub-area images includes the following steps:

[0074] S11, in each sub-area, an infrared thermal image in the sub-area is acquired by an infrared image acquisition device, and a color image in the sub-area is acquired by a color image acquisition card;

[0075] S12. According to the characteristics of the infrared thermal image, the mean method is used to extract the features of the temperature rise area to obtain the source image feature map; the mean method is to calculate the mean pixel value of the entire infrared thermal image. Compare the mean of the region of interest with the mean of the entire image. If the mean of a certain area is significantly higher than the mean of the entire image, this area may be a temperature rise area. Generate a binary feature map. On this feature map, the temperature rise area is marked as a high value, and the rest of the area is marked as a low value. In order to remove possible noise and enhance the features, the feature map can be filtered, such as median filtering.

[0076] S13, using the source image feature map to calculate the gradient field fusion weight, and using the multi-scale transformation of Gaussian filtering combined with the image gradient covariance to obtain the structural features of the infrared thermal image;

[0077] S14, select the scale factor;

[0078] It should be noted that Gaussian filtering is a method of smoothing images, which can reduce image noise and details and highlight the main structure of the image. Different scale factors can control the degree of smoothing. The larger the scale factor, the stronger the smoothing and the more blurred the image. Image gradient covariance is a matrix that describes the local rate of change of an image. It can reflect the edges, corners and other features of the image. The eigenvalues ​​and eigenvectors of the image gradient covariance can represent the maximum and minimum rate of change of the image at a certain point and its direction. Combining the multi-scale transformation of Gaussian filtering with the image gradient covariance can obtain the image structure features at different scales. In this way, according to different fusion purposes, the appropriate scale factor can be selected so that the fusion result can retain the important information of the source image.

[0079] S15, using the soft-max function to soften the gradient weights of the edge areas of the infrared thermal image, and obtaining a structure tensor with improved gradient weights, so that the infrared and color images can be fused more smoothly during the fusion process and possible artifacts can be reduced;

[0080] S16. Reconstruct the image based on the variational technique and the Poisson equation solution; wherein the variational technique is used for image denoising. For example, the Rudin-Osher-Fatemi (ROF) model is a classic variational denoising method. The Poisson equation refers to the Poisson fusion of the image, which maintains the gradient of the source image in the target image while making the fused result as close to the target image as possible. To this end, the following equation is used: Among them, f is the required fusion result, s is the source image, and is the gradient operator. This equation means that the gradient of the source image is maintained in the result image. To solve this equation, Poisson's equation is used: Δf = Δs. Where Δ is the Laplace operator.

[0081] The reconstructed image is obtained by solving the Poisson equation (eg, using an iterative technique such as the Gauss-Seidel method or the Jacobi method).

[0082] S17. Use transparency method to fuse the color image and the reconstructed image to obtain the sub-region image; transparency method adjusts the opacity of the two images so that both can be displayed in the result; transparency method (or alpha blending) is used to fuse two or more images. The basic idea is to give each image a weight (also called alpha value) for each pixel, and then perform weighted fusion.

[0083] S18, extracting features of the sub-region image using image processing technology, including texture, shape, and temperature features, and establishing historical data based on the features;

[0084] S19. Establish a deep learning model, and use historical data to train the deep learning model, input the features of the new sub-region image into the trained deep learning model, and detect risk factors.

[0085] In this embodiment, when the gradient field fusion weight is calculated using the source image feature map, the gradient field fusion weight is:

[0086]

[0087] In the formula, S n Represents the feature map of the nth source image;

[0088] S k Represents the feature map of the k-th source image;

[0089] N represents the total number of source images.

[0090] In this embodiment, the step of selecting a scale factor includes the following steps:

[0091] S141. Comparative analysis of multi-scale factors on infrared thermal images;

[0092] S142: Select a scale factor according to the richness of image detail information.

[0093] S2. A smart wearable device is configured on the wearer, and a sensor is configured on the smart wearable device to collect location data.

[0094] S3. Plan several walking routes based on the activity area and recommend walking paths to the wearer in real time. If dangerous factors appear on the wearer's walking path, issue an early warning and recommend a new walking path.

[0095] In a further embodiment, the planning of several walking routes based on the activity area and the recommendation of walking paths to the wearer in real time, and if a risk factor appears on the wearer's walking path, an early warning is issued and a new walking path is recommended, including the following steps:

[0096] S31, taking each sub-area of ​​the activity area as a node of the graph, taking the road between any two nodes as an edge of the graph, and assigning weights to the edges according to the distance between the two points, the estimated travel time, and the congestion level of the road;

[0097] S32, update edge weights in real time;

[0098] S33, according to the current graph weight and based on the genetic algorithm, obtaining the optimal path from the starting point to the end point;

[0099] S34, if the optimal path is determined, the optimal path is sent to the wearer's smart wearable device in real time by wireless transmission, and the wearer is prompted to walk along the optimal path;

[0100] S35, if the wearer deviates from the optimal path, the optimal path is recalculated and updated and pushed to the wearer;

[0101] S36. If a dangerous factor appears on any edge of the wearer's walking path, an early warning is issued, the weight of the edge is reduced, the optimal path is recalculated, and the update is pushed to the wearer.

[0102] In this embodiment, the real-time updating of edge weights includes the following steps:

[0103] S321. Collect road information in the activity area in real time and classify the road information into positive information and negative information; positive information includes smooth traffic, good safety conditions, comfortable environment, etc., and negative information includes blocked traffic, safety hazards, unsuitable environment, etc. In order to achieve classification, various data sources can be combined, including:

[0104] Sensor data: For example, cameras, infrared sensors, sound sensors, etc., can provide real-time environmental information.

[0105] Public data sources: such as weather forecasts, traffic updates, event schedules, etc.

[0106] User Feedback: Pedestrians or vehicle drivers can provide real-time feedback through the app to report problems or obstacles they encounter.

[0107] S322. If negative information appears, the weight of the corresponding edge in the graph is increased, and the cost increases as the weight increases.

[0108] In this embodiment, the step of obtaining the optimal path from the starting point to the end point based on the current graph weight and the genetic algorithm includes the following steps:

[0109] S331, encoding the walking path into chromosomes, and each chromosome represents a path;

[0110] S332, randomly generate a number of chromosomes and form a primary population;

[0111] S333, calculating the total cost of each path according to the weight of the edge in the graph;

[0112] S334, evaluating each chromosome according to the total cost through the fitness function, and the lower the total cost, the higher the fitness; wherein the diversity of the population can be increased by crossover or mutation;

[0113] S335, setting a termination condition; specifically, a variety of termination conditions may be set, such as reaching a preset number of iterations, fitness reaching a certain threshold, fitness not improving after multiple consecutive iterations, etc.;

[0114] S336: If the termination condition is triggered, the chromosome with the highest fitness is selected and decoded into a path, and pushed to the wearer's smart wearable device.

[0115] In this embodiment, when each chromosome is evaluated according to the total cost through the fitness function, the fitness function is:

[0116]

[0117] In the formula, D represents the total cost of each path;

[0118] ε represents a positive number that is divisible by zero.

[0119] S4. If the wearer approaches or enters a sub-area where dangerous factors occur, an alarm is issued through the smart wearable device.

[0120] In a further embodiment, if the wearer enters a sub-area where a dangerous factor occurs, the alarm is issued by the smart wearable device, including the following steps:

[0121] S41, determining the time length between the time when the risk factor appears and the current time, if the time length is less than a preset threshold, determining that the risk factor is valid;

[0122] S42. If the risk factor is valid, an alarm is issued through the smart wearable device.

[0123] like Figure 2 As shown, according to another embodiment of the present invention, a danger perception system for smart wearable devices based on infrared technology is provided, which includes an image acquisition module 1, a position data acquisition module 2, a path warning module 3 and an alarm module 4.

[0124] Wherein, the image acquisition module 1 and the location data acquisition module 2 are both connected to the path warning module 3 and the alarm module 4 .

[0125] The image acquisition module 1 is used to configure a plurality of infrared image acquisition devices and color image acquisition cards in the activity area, and to acquire sub-area images through the infrared image acquisition devices and color image acquisition cards in each sub-area, and to identify risk factors in the sub-area images.

[0126] The position data collection module 2 is used to configure a smart wearable device on the wearer, and configure a sensor on the smart wearable device to collect position data.

[0127] The path warning module 3 is used to plan several walking routes based on the activity area and recommend walking routes to the wearer in real time. If a dangerous factor appears on the wearer's walking path, an early warning is issued and a walking route is re-recommended.

[0128] The alarm module 4 is used to issue an alarm through the smart wearable device if the wearer enters a sub-area where dangerous factors appear.

[0129] In summary, the present invention provides a method and system for intelligent wearable device danger perception based on infrared technology, which solves the safety problem of walking in a wild night environment, especially for those possible environmental hazards. The present invention combines infrared technology with color image technology, which can not only capture images in the dark, but also identify potential dangers based on temperature differences and infrared characteristics, and fuse the color image with the reconstructed image to obtain a sub-region image, so that each sub-region image makes up for the problem of lack of detailed information in a simple infrared thermal image, which is particularly useful in a wild environment and can identify animals, protruding obstacles or other heat sources. The present invention can dynamically recommend the safest and fastest path to the user, and re-plan the route in time when danger occurs to ensure that the user always walks on the safest path, and by using a genetic algorithm, the optimal solution can be found quickly and efficiently in multiple possible paths, which is of high value in a changing environment. When the user approaches or enters a dangerous area, the intelligent wearable device will provide a timely warning to help the user avoid potential risks.

[0130] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for hazard perception of smart wearable devices based on infrared technology, characterized in that: The method comprises the following steps: S1. Several infrared image acquisition devices and color image acquisition cards are configured in the activity area, and sub-area images are acquired by the infrared image acquisition devices and color image acquisition cards in each sub-area, and risk factors are identified in the sub-area images; The method of configuring a plurality of infrared image acquisition devices and color image acquisition cards in the activity area, and acquiring sub-area images in each sub-area by using the infrared image acquisition devices and color image acquisition cards, and identifying risk factors in the sub-area images includes the following steps: S11, in each sub-area, an infrared thermal image in the sub-area is acquired by an infrared image acquisition device, and a color image in the sub-area is acquired by a color image acquisition card; S12. According to the characteristics of the infrared thermal image, the mean value method is used to extract the features of the temperature rise area to obtain a source image feature map; S13, using the source image feature map to calculate the gradient field fusion weight, and using the multi-scale transformation of Gaussian filtering combined with the image gradient covariance to obtain the structural features of the infrared thermal image; S14, select the scale factor; S15, using a soft-max function to soften the gradient weight of the edge area of ​​the infrared thermal image, and obtaining a structure tensor with improved gradient weight; S16, obtaining a reconstructed image according to the variational technique and the Poisson equation solution; S17, using a transparency method to fuse the color image and the reconstructed image to obtain a sub-region image; S18, extracting features of the sub-region image using image processing technology, and establishing historical data based on the features; S19, establishing a deep learning model, and using historical data to train the deep learning model, and inputting the features of the new sub-region image into the trained deep learning model, and performing risk factor detection; The step of selecting the scale factor comprises the following steps: S141. Comparative analysis of multi-scale factors on infrared thermal images; S142, selecting a scale factor according to the richness of image detail information; S2, configuring a smart wearable device on the wearer, and configuring a sensor on the smart wearable device to collect location data; S3. Plan several walking routes based on the activity area and recommend walking routes to the wearer in real time. If a dangerous factor appears on the wearer's walking route, issue an early warning and re-recommend a walking route; S4. If the wearer enters a sub-area where dangerous factors occur, an alarm is issued through the smart wearable device; The method of planning a plurality of walking routes based on the activity area, and recommending walking routes to the wearer in real time, and issuing an early warning if a risk factor appears on the wearer's walking route, and re-recommending a walking route includes the following steps: S31, taking each sub-area of ​​the activity area as a node of the graph, taking the road between any two nodes as an edge of the graph, and assigning weights to the edges according to the distance between the two points, the estimated travel time, and the congestion level of the road; S32, update edge weights in real time; S33, according to the current graph weight and based on the genetic algorithm, obtaining the optimal path from the starting point to the end point; S34, if the optimal path is determined, the optimal path is sent to the wearer's smart wearable device in real time by wireless transmission, and the wearer is prompted to walk along the optimal path; S35, if the wearer deviates from the optimal path, the optimal path is recalculated and updated and pushed to the wearer; S36. If a risk factor appears on any edge of the wearer's walking path, an early warning is issued, the weight of the edge is reduced, and the optimal path is recalculated and updated and pushed to the wearer; The method of obtaining the optimal path from the starting point to the end point based on the current graph weight and the genetic algorithm comprises the following steps: S331, encoding the walking path into chromosomes, and each chromosome represents a path; S332, randomly generate a number of chromosomes and form a primary population; S333, calculating the total cost of each path according to the weight of the edge in the graph; S334, evaluating each chromosome according to the total cost through the fitness function, and the lower the total cost, the higher the fitness; S335, setting a termination condition; S336: If the termination condition is triggered, the chromosome with the highest fitness is selected and decoded into a path, and pushed to the wearer's smart wearable device at the same time; When each chromosome is evaluated according to the total cost through the fitness function, the fitness function is: In the formula, D represents the total cost of each path; Represents a positive number that is divisible by zero.

2. According to the infrared technology-based intelligent wearable device danger perception method of claim 1, it is characterized in that: When the gradient field fusion weight is calculated using the source image feature map, the gradient field fusion weight is: In the formula, S n Indicates n feature maps of source images; S k Indicates k feature maps of source images; N Indicates the total number of source images.

3. According to the infrared technology-based intelligent wearable device danger perception method of claim 1, it is characterized in that: The real-time updating of edge weights includes the following steps: S321, collecting road information in the activity area in real time, and dividing the road information into positive information and negative information; S322. If negative information appears, the weight of the corresponding edge in the graph is increased, and the cost increases as the weight increases.

4. According to the infrared technology-based intelligent wearable device danger perception method of claim 1, it is characterized in that: If the wearer enters a sub-area where dangerous factors occur, the smart wearable device issues an alarm, which includes the following steps: S41, determining the time length between the time when the risk factor appears and the current time, if the time length is less than a preset threshold, determining that the risk factor is valid; S42. If the risk factor is valid, an alarm is issued through the smart wearable device.

5. A smart wearable device danger perception system based on infrared technology, used to implement the smart wearable device danger perception method based on infrared technology according to any one of claims 1 to 4, characterized in that: The system includes an image acquisition module, a location data acquisition module, a path warning module and an alarm module; Wherein, the image acquisition module and the location data acquisition module are both connected to the path warning module and the alarm module; The image acquisition module is used to configure a plurality of infrared image acquisition devices and color image acquisition cards in the activity area, and to acquire sub-area images in each sub-area through the infrared image acquisition devices and color image acquisition cards, and to identify risk factors in the sub-area images; The position data collection module is used to configure a smart wearable device on the wearer and configure a sensor on the smart wearable device to collect position data; The path warning module is used to plan several walking routes based on the activity area and recommend walking routes to the wearer in real time. If a dangerous factor appears on the wearer's walking route, it will issue a warning and re-recommend the walking route; The alarm module is used to issue an alarm through the smart wearable device if the wearer enters a sub-area where dangerous factors appear.