Method for predicting positions of lost persons under mountain torrent and debris flow disasters

Abstract

The invention provides a method for predicting positions of lost persons under mountain torrent and debris flow disasters, and belongs to the field of disaster rescue. The method comprises the following steps: acquiring multi-source data fusing a thermal infrared image, a radar image and a terrain; feature extraction is carried out on the acquired multi-source data, so that the positions of the lost persons under the mountain torrent and debris flow disasters can be identified in a multi-source mode from person tracks, human body heat radiation and human body skeleton detection; constructing a comprehensive analysis function based on the extracted features; and based on the comprehensive analysis function, obtaining predicted values of different positions at the time t, and completing prediction of the position of the lost person. The invention aims to solve the problems of low accuracy and low efficiency of the existing lost person position prediction method.

Description

Technical Field

[0001] The present invention belongs to the field of disaster relief, and in particular relates to a method for predicting the location of missing persons in flash flood and mud-rock flow disasters. Background Art

[0002] Flash floods and mudslides are common natural disasters. Due to their suddenness and destructive nature, they often cause people to go missing. Rapidly locating missing individuals is crucial to rescue efforts. Currently, rescue teams primarily conduct search and rescue operations through a combination of manual searches and drone inspections. Life detectors have also become a crucial tool in disaster relief efforts. However, traditional manual ground searches are significantly restricted by terrain and face significant safety threats from secondary disasters. Traditional drone aerial searches only retrieve surface information and are significantly affected by weather and visibility. Life detectors are only sensitive to individuals with vital signs and often reflect weaker signals from buried individuals who are stationary and whose vital signs are very weak. Due to the complex terrain and harsh environment at disaster sites, traditional search and rescue methods are often inefficient, inaccurate, and have limited coverage. Furthermore, missing individuals may no longer be showing vital signs, posing significant challenges to search and rescue efforts. Summary of the Invention

[0003] In response to the above-mentioned deficiencies in the prior art, the present invention provides a method for predicting the location of missing persons in flash flood and mud-rock flow disasters, aiming to solve the problems of low accuracy and low efficiency of the existing methods for predicting the location of missing persons.

[0004] In order to achieve the above objectives, the present invention adopts a technical solution: a method for predicting the location of missing persons in a flash flood and mud-rock flow disaster, comprising the following steps:

[0005] Acquire multi-source data that fuses thermal infrared images, radar images, and terrain;

[0006] Feature extraction is performed on the acquired multi-source data to identify the location of missing persons in flash flood and mud-rock flow disasters through multiple sources including person trajectory, human thermal radiation, and human skeleton detection.

[0007] Based on the extracted features, a comprehensive analysis function is constructed;

[0008] Based on the comprehensive analysis function, the predicted values ​​of different locations at time t are obtained to complete the prediction of the location of the missing person.

[0009] The beneficial effects of the present invention are: the present invention establishes a personnel trajectory model by fusing multi-source data, combines multi-source data fusion technology, and comprehensively considers factors such as thermal infrared data and radar image data to achieve accurate prediction of the location of missing persons. Even when the missing persons have no vital signs, it can play an important role and provide effective technical support for disaster relief work.

[0010] Furthermore, the acquisition of multi-source data is specifically as follows:

[0011] Aerial surveys are conducted using drones equipped with a positioning attitude system, optical cameras, thermal infrared sensors, and ground-penetrating radar equipment. By acquiring the positioning attitude system POS data, ground object reflection intensity, target object radiation intensity, and underground echo reflection signals, terrain data, thermal infrared image data, and radar image data are obtained.

[0012] The beneficial effect of the above further scheme is that the present invention innovatively integrates thermal infrared data, radar image data and personnel trajectory data (personnel trajectory data is obtained from the personnel trajectory model, the data source is terrain data, and terrain data is calculated by combining POS data with optical image stereo relative calculation), obtains information from different levels, and comprehensively analyzes the possible location of the missing person, which is the core of achieving accurate prediction, that is, integrating thermal infrared, radar and personnel trajectory multi-source data to accurately determine the location of the missing person from different dimensions. The present invention effectively improves the prediction accuracy by complementing each other with various data, provides accurate direction for rescue, avoids blind search, and improves rescue efficiency.

[0013] Furthermore, the feature extraction includes:

[0014] Temperature anomaly extraction: Extract temperature anomaly areas in thermal infrared images and perform standardization processing:

[0015]

[0016] Where T(x,y,t) represents the radiation intensity detected by the thermal infrared sensor at position (x,y) and time t, I(x,y,t) represents the radiation intensity of the human body, and μ T (t) represents the average radiation intensity of the thermal infrared image at time t, σ T (t) represents the standard deviation of the radiation intensity of the thermal infrared image at time t, t0 represents the disaster start time, and τ represents the radiation decay time constant;

[0017] Ground penetration underground echo reflection signal analysis: Analyze the underground echo reflection signals of human bones in radar image data and enhance the recognition of underground echo reflection signals of human bones through standardization processing:

[0018]

[0019] Among them, R(x,y,t) represents the radar skeleton underground echo reflection signal strength received by the ground penetrating radar at position (x,y) and time t, S(x,y,t) represents the underground echo reflection signal, μ R(t) represents the average echo intensity of the radar image data at time t, d(x, y) represents the buried depth, v represents the electromagnetic wave propagation velocity, α(ξ) represents the soil medium attenuation coefficient, dξ represents the differential of the integral variable, Δt(x, y, t) represents the time difference between the electromagnetic wave transmitted by the ground penetrating radar at position (x, y) and time t and the reflected echo received;

[0020] Personnel trajectory analysis: Based on the positioning posture system POS data and optical image stereo relative modeling, terrain data is obtained. Through terrain analysis, the slope G(x, y), slope direction D(x, y), and the difference Δt between the disaster occurrence time t and the disaster start time t0 are considered to construct the personnel trajectory model formula:

[0021]

[0022] Among them, P(x, y, t) represents the probability distribution of the personnel trajectory model, σ represents the weight parameter that controls the influence of terrain factors on the probability, θ0 represents the reference slope value, τ P represents the time-dependent decay parameter.

[0023] The beneficial effects of the above further scheme are: targeted processing and feature extraction of various types of data, such as temperature anomaly extraction of thermal infrared data, underground echo reflection signal analysis of radar image data, and model construction of personnel trajectory, laying the foundation for subsequent fusion analysis.

[0024] Furthermore, the expression of the comprehensive analysis function is as follows:

[0025] F(x,y,t)=ω1T(x,y,t)+ω2R(x,y,t)+ω3P(x,y,t)

[0026]

[0027] Among them, F(x,y,t) represents the predicted value at position (x,y) and time t, T(x,y,t) represents the radiation intensity detected by the thermal infrared sensor at position (x,y) and time t, R(x,y,t) represents the intensity of the radar bone underground echo reflection signal received by the ground penetrating radar at position (x,y) and time t, P(x,y,t) represents the probability distribution of the personnel trajectory model, ω1, ω2 and ω3 all represent weight coefficients, d(x,y) represents the burial depth, d avg It represents the average buried depth of all detection points, d max and d min They respectively represent the maximum and minimum values ​​of the buried depth of all detection points.

[0028] The beneficial effect of the above further solution is that the present invention determines the weight of the comprehensive analysis function based on determinable factors such as the buried depth, reasonably balances the role of different data sources in the prediction, and improves the prediction accuracy.

[0029] Furthermore, based on the comprehensive analysis function, the predicted values ​​of different positions at time t are obtained, which are specifically:

[0030] Set the judgment threshold F threshold , when the comprehensive analysis function F(x,y,t) is greater than the judgment threshold F threshold , there is a missing person at that location.

[0031] The beneficial effect of the above further scheme is: by comparing the predicted value of the comprehensive analysis function with the threshold, key areas can be quickly and accurately screened to enhance the targeted rescue. The threshold is set based on empirical science and can be flexibly adjusted to adapt to different disaster scenarios, thus buying time for rescue. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 Flow chart of the method of the present invention.

[0033] Figure 2 Schematic diagram of the method framework of the present invention. DETAILED DESCRIPTION

[0034] The specific embodiments of the present invention are described below to facilitate understanding of the present invention by those skilled in the art. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the appended claims, these changes are obvious, and all inventions and creations utilizing the concepts of the present invention are protected.

[0035] Example

[0036] like Figure 1 and Figure 2 As shown, the present invention provides a method for predicting the location of missing persons in a flash flood and mud-rock flow disaster, and the implementation method is as follows:

[0037] Acquire multi-source data integrating thermal infrared images, radar images, and terrain, specifically:

[0038] Aerial surveys are conducted using drones equipped with a positioning attitude system, optical cameras, thermal infrared sensors, and ground-penetrating radar equipment. By acquiring the positioning attitude system POS data, ground object reflection intensity, target object radiation intensity, and underground echo reflection signals, terrain data, thermal infrared image data, and radar image data are obtained.

[0039] In this embodiment, an unmanned aerial vehicle (UAV) equipped with a positioning attitude system (POS), an optical camera, a thermal infrared sensor, and a ground-penetrating radar device is used for aerial surveying to obtain the positioning attitude system POS data, the reflection intensity of the ground objects, the radiation intensity of the target objects, and the underground echo reflection signal, thereby obtaining terrain data, thermal infrared images, and radar image data.

[0040] UAV aerial survey: Obtain optical image data and the UAV's positioning attitude system POS data, including position and attitude information, to provide a spatial positioning basis for subsequent data processing.

[0041] Thermal infrared detection: This technology uses infrared detection lenses (or thermal infrared sensors) to capture infrared radiation emitted by the human body and obtain the radiation intensity value of the target object. Even if a person has no vital signs, they will still emit infrared radiation before they become completely hypothermic, which is helpful for detecting missing persons.

[0042] Ground-penetrating radar detection. Based on the differences in electromagnetic wave properties of different materials, high-frequency electromagnetic waves are emitted into the ground. When they encounter human bones, they reflect strong underground echo signals. The target location is determined by the time difference of the underground echo reflection signal and the propagation speed of the electromagnetic wave.

[0043] Feature extraction is performed on the acquired multi-source data to identify the location of missing persons in flash flood and mud-rock flow disasters through multiple sources including person trajectory, human thermal radiation, and human skeleton detection.

[0044] In this embodiment, temperature anomaly extraction: extract the temperature anomaly area in the thermal infrared image, and perform normalization processing to highlight the temperature anomaly area:

[0045]

[0046] Where T(x,y,t) represents the radiation intensity detected by the thermal infrared sensor at position (x,y) and time t, I(x,y,t) represents the radiation intensity of the human body, and μ T (t) represents the average radiation intensity of the thermal infrared image at time t, σ T (t) represents the standard deviation of the radiation intensity of the thermal infrared image at time t, t0 represents the disaster start time, and τ represents the radiation decay time constant (the empirical value is 2 to 6 hours).

[0047] Ground penetration underground echo reflection signal analysis: Analyze the underground echo reflection signals of human bones in radar image data and enhance the recognition ability of underground echo reflection signals of human bones through standardized processing:

[0048]

[0049] Among them, R(x,y,t) represents the radar skeleton underground echo reflection signal strength received by the ground penetrating radar at position (x,y) and time t, S(x,y,t) represents the underground echo reflection signal, μ R (t) represents the average echo intensity of the radar image data at time t, d(x,y) represents the buried depth, v represents the propagation velocity of the electromagnetic wave, α(ξ) represents the attenuation coefficient of the soil medium, dξ represents the differential of the integral variable, and represents the small displacement increment on the path. Δt(x,y,t) represents the time difference between the electromagnetic wave transmitted by the ground penetrating radar at position (x,y) and time t and the reflected echo received.

[0050] Personnel trajectory analysis: Based on the positioning posture system POS data, by analyzing the terrain (such as slope, slope direction, etc.), considering the slope G(x, y), slope direction D(x, y), and the difference Δt = t-t0 between the disaster occurrence time t and the disaster start time t0, a personnel trajectory model formula is constructed:

[0051]

[0052] Where P(x, y, t) represents the probability distribution of the personnel trajectory model, θ0 represents a reference slope value related to the common escape direction (which can be set based on experience. For example, in mountainous areas, it is generally believed that the common escape direction is down the valley, and the slope of this direction can be set as θ0), σ represents the weight parameter that controls the influence of terrain factors on probability, and τ P Represents a time-dependent error reduction parameter (which can be set based on the disaster type and past experience. For example, in flash floods and mudslides, it can be set to 1-3 hours, indicating that the probability of a person remaining in the same location or moving along their original trajectory decreases over time). This formula uses an exponential function to comprehensively consider the impact of terrain and time on the probability of a person's possible location, and does not rely on statistical analysis of historical data.

[0053] Based on the extracted features, a comprehensive analysis function is constructed;

[0054] In this embodiment, a comprehensive analysis function is constructed to comprehensively consider the personnel trajectory model, thermal infrared data, and ground penetrating radar image data factors, perform weighted superposition, and establish a personnel location model. The comprehensive analysis function is defined as:

[0055] F(x,y,t)=ω1T(x,y,t)+ω2R(x,y,t)+ω3P(x,y,t)

[0056] Considering the influence of the buried depth d(x,y) on the reliability of different detection data, the weight is determined as follows:

[0057]

[0058] Among them, F(x,y,t) represents the predicted value at position (x,y) and time t, T(x,y,t) represents the radiation intensity detected by the thermal infrared sensor at position (x,y) and time t, R(x,y,t) represents the intensity of the radar bone underground echo reflection signal received by the ground penetrating radar at position (x,y) and time t, P(x,y,t) represents the probability distribution of the personnel trajectory model, ω1, ω2 and ω3 all represent weight coefficients, d(x,y) represents the burial depth, d avg It represents the average buried depth of all detection points, d max and d min They respectively represent the maximum and minimum values ​​of the buried depth of all detection points.

[0059] This weight determination method is based on the determinable factor of burial depth. As the burial depth increases, thermal infrared data is more affected and its weight decreases; radar image data is relatively more reliable and its weight increases; the weight of the personnel trajectory model comprehensively considers the relationship between the burial depth and the average burial depth to balance the effects of various data at different depths.

[0060] Based on the comprehensive analysis function, the predicted values ​​of different locations at time t are obtained to complete the prediction of the location of the missing person, which is specifically as follows:

[0061] Set the judgment threshold F threshold , when the comprehensive analysis function F(x,y,t) is greater than the judgment threshold F threshold , there is a missing person at that location.

[0062] In this embodiment, the present invention combines optical images captured by drone aerial surveys with markers on the images that meet the criteria, generating a final thematic map of the predicted results. Rescuers can use the marked locations on the thematic map to quickly locate areas where missing persons may be located, improving rescue efficiency.

[0063] In summary, based on the above design, the beneficial effects of the present invention include:

[0064] Enhanced Precision Positioning: This system integrates data from multiple sources, including thermal infrared sensors, ground-penetrating radar, and personnel trajectory data, to accurately determine the location of missing personnel from multiple dimensions. These data complement each other, effectively improving prediction accuracy and providing precise direction for rescue operations, avoiding blind searches and increasing rescue efficiency.

[0065] Adaptability to complex environments: Thermal infrared is not affected by light, radar is not afraid of obstructions, and personnel trajectories take terrain and time into consideration, enabling the method to adapt to the complex environment after mountain torrents and mudslides, overcoming the limitations of traditional means, broadening the scope of application, and enhancing rescue reliability.

[0066] Highly efficient response: Utilizing efficient data processing technology and a specific weighting method, prediction results are quickly generated. This helps rescuers seize the golden rescue window, increase the chances of rescuing missing individuals, and create more opportunities for lifesaving efforts.

[0067] Optimized resource allocation: Accurate predictions enable more rational allocation of rescue resources, avoiding the waste of resources caused by the blindness of traditional search and rescue operations. This not only improves resource utilization and reduces costs, but also ensures the safety of rescuers.

[0068] Excellent expansion and compatibility: The present invention has good scalability and compatibility, is easy to integrate new data sources and processing methods, and the equipment involved is universal, which makes it easy to combine with the existing rescue system and facilitate large-scale promotion and application.

Claims

1. A method for predicting the location of missing persons in a flash flood and mud-rock flow disaster, characterized in that: The following steps are involved: Acquire multi-source data that fuses thermal infrared images, radar images, and terrain; Feature extraction is performed on the acquired multi-source data to identify the location of missing persons in flash flood and mud-rock flow disasters through multiple sources including person trajectory, human thermal radiation, and human skeleton detection. Based on the extracted features, a comprehensive analysis function is constructed; Based on the comprehensive analysis function, the predicted values ​​of different locations at time t are obtained to complete the prediction of the location of the missing person.

2. The method for predicting the location of missing persons in a flash flood and mud-rock flow disaster according to claim 1, characterized in that: The acquisition of multi-source data is specifically as follows: Aerial surveys are conducted using drones equipped with a positioning attitude system, optical cameras, thermal infrared sensors, and ground-penetrating radar equipment. By acquiring the positioning attitude system POS data, ground object reflection intensity, target object radiation intensity, and underground echo reflection signals, terrain data, thermal infrared image data, and radar image data are obtained.

3. The method for predicting the location of missing persons in a flash flood and mud-rock flow disaster according to claim 1, characterized in that: The feature extraction includes: Temperature anomaly extraction: Extract temperature anomaly areas in thermal infrared images and perform standardization processing: Where T(x,y,t) represents the radiation intensity detected by the thermal infrared sensor at position (x,y) and time t, I(x,y,t) represents the radiation intensity of the human body, and μ T (t) represents the average radiation intensity of the thermal infrared image at time t, σ T (t) represents the standard deviation of the radiation intensity of the thermal infrared image at time t, t0 represents the disaster start time, and τ represents the radiation decay time constant; Ground penetration underground echo reflection signal analysis: Analyze the underground echo reflection signals of human bones in radar image data and enhance the recognition of underground echo reflection signals of human bones through standardization processing: Among them, R(x,y,t) represents the radar skeleton underground echo reflection signal strength received by the ground penetrating radar at position (x,y) and time t, S(x,y,t) represents the underground echo reflection signal, μ R (t) represents the average echo intensity of the radar image data at time t, d(x, y) represents the buried depth, v represents the electromagnetic wave propagation velocity, α(ξ) represents the soil medium attenuation coefficient, dξ represents the differential of the integral variable, Δt(x, y, t) represents the time difference between the electromagnetic wave transmitted by the ground penetrating radar at position (x, y) and time t and the reflected echo received; Personnel trajectory analysis: Based on the positioning posture system POS data and optical image stereo relative modeling, terrain data is obtained. Through terrain analysis, the slope G(x, y), slope direction D(x, y), and the difference Δt between the disaster occurrence time t and the disaster start time t0 are considered to construct the personnel trajectory model formula: Among them, P(x, y, t) represents the probability distribution of the personnel trajectory model, σ represents the weight parameter that controls the influence of terrain factors on the probability, θ0 represents the reference slope value, τ P represents the time-dependent decay parameter.

4. The method for predicting the location of missing persons in a flash flood and mud-rock flow disaster according to claim 1, characterized in that: The expression of the comprehensive analysis function is as follows: F(x,y,t)=ω1T(x,y,t)+ω2R(x,y,t)+ω3P(x,y,t) Among them, F(x,y,t) represents the predicted value at position (x,y) and time t, T(x,y,t) represents the radiation intensity detected by the thermal infrared sensor at position (x,y) and time t, R(x,y,t) represents the intensity of the radar bone underground echo reflection signal received by the ground penetrating radar at position (x,y) and time t, P(x,y,t) represents the probability distribution of the personnel trajectory model, ω1, ω2 and ω3 all represent weight coefficients, d(x,y) represents the burial depth, d avg It represents the average buried depth of all detection points, d max and d min They respectively represent the maximum and minimum values ​​of the buried depth of all detection points.

5. The method for predicting the location of missing persons in a flash flood and mud-rock flow disaster according to claim 1, characterized in that: Based on the comprehensive analysis function, the predicted values ​​of different positions at time t are obtained, which are specifically: Set the judgment threshold F threshold , when the comprehensive analysis function F(x,y,t) is greater than the judgment threshold F threshold , there is a missing person at that location.

Images