Infrared detection-based external wall hollowing and falling risk prediction method and system

Abstract

The embodiment of the application provides a kind of based on infrared detection's outer wall hollowing out and falling risk prediction method and system, the method comprises: based on preset time interval, the wall material of the building outer wall to be detected at different time points is acquired;And for the wall material of each time point in different time points, from the mapping relationship library determined in advance, obtain the multiple sampling distances corresponding to wall material;Based on multiple sampling distances, the building outer wall to be detected is physically sampled, and multiple distance sampling images are generated;Based on multiple distance sampling images, image fusion is carried out, and the wall temperature field image of the building outer wall to be detected at different time points is generated;And based on the wall temperature field image of different time points, determine temperature field image sequence;Based on temperature field image sequence, outer wall risk detection and risk level division are carried out, and the current risk level of the building outer wall to be detected is determined.The above scheme can improve the accuracy of outer wall hollowing out and falling risk prediction.

Description

Technical Field

[0001] This application relates to the field of building technology, specifically to a method and system for predicting the risk of hollowing and falling off of exterior walls based on infrared detection. Background Technology

[0002] As urban buildings age, incidents of hollowing and detachment of exterior wall cladding bricks, insulation layers, and other components are frequent, posing a significant threat to public safety. Currently, the detection of hollow exterior walls mainly relies on manual visual inspection, tapping, and infrared thermal imaging technology. Among these, infrared thermal imaging, with its advantages of being non-contact and providing rapid coverage, is gradually becoming the mainstream detection method. However, it still has significant limitations when applied to complex and large-scale building exteriors.

[0003] Existing infrared detection methods using drones or robots often employ fixed-distance or single-height sampling. However, the spatial resolution of infrared thermal imagers is directly related to the shooting distance, leading to a dilemma: "high resolution at close range but low efficiency and prone to registration errors, while high efficiency at long range but missing small hollow areas." Furthermore, the thermal physical properties of wall materials (tiles, paints, etc.) vary greatly, and a uniform sampling distance cannot capture the optimal thermal contrast, further increasing the risk of misjudgment and missed detection. This results in insufficient accuracy and reliability in predicting the risk of hollow areas and detachment of exterior walls. Summary of the Invention

[0004] This application aims to provide a method and system for predicting the risk of hollow and detached exterior walls based on infrared detection, which can improve the accuracy of predicting the risk of hollow and detached exterior walls.

[0005] The technical solution of this application is implemented as follows:

[0006] In a first aspect, embodiments of this application provide a method for predicting the risk of hollowing and detachment of exterior walls based on infrared detection, the method comprising:

[0007] Based on a preset time interval, the wall material of the building exterior wall to be detected at different time points is obtained; and for the wall material at each of the different time points, multiple sampling distances corresponding to the wall material are obtained from a preset mapping relationship library; wherein, the mapping relationship library is a mapping relationship between the wall material and multiple sampling distances;

[0008] Based on the multiple sampling distances, physical sampling is performed on the exterior wall of the building to be detected to generate multi-distance sampling images;

[0009] Based on the multi-distance sampled images, image fusion is performed to generate wall temperature field images of the building exterior wall at different time points; and based on the wall temperature field images at different time points, a temperature field image sequence is determined.

[0010] Based on the temperature field image sequence, external wall risk detection and risk level classification are performed to determine the current risk level of the external wall of the building to be detected.

[0011] In the above scheme, the step of physically sampling the exterior wall of the building to be detected based on the multiple sampling distances to generate a multi-distance sampling image includes:

[0012] Determine the physical dimensions and target resolution of the exterior wall of the building to be inspected; and determine the image pixel size based on the physical dimensions and target resolution.

[0013] Determine the effective rectangular region resolution corresponding to each of the plurality of sampling distances, and the sampling interval corresponding to each of the plurality of sampling distances; wherein, the sampling interval includes a horizontal interval and a vertical interval;

[0014] Based on the multiple sampling distances, the image pixel size, the effective rectangular area resolution, and the sampling interval, physical sampling is performed on the exterior wall of the building to be detected to generate a multi-distance sampling image.

[0015] In the above scheme, the step of physically sampling the exterior wall of the building to be detected based on the multiple sampling distances, the image pixel size, the effective rectangular region resolution, and the sampling interval to generate a multi-distance sampling image includes:

[0016] Based on the image pixel size and the effective rectangular region resolution, the number of effective rectangular regions corresponding to each of the multiple sampling distances is calculated; wherein, the number of effective rectangular regions includes the number of horizontal effective rectangles and the number of vertical effective rectangles;

[0017] For each of the plurality of sampling distances, the horizontal sampling index and the vertical sampling index of each sampling distance are calculated based on the number of horizontal valid rectangles, the number of vertical valid rectangles, the horizontal interval, and the vertical interval.

[0018] Based on the horizontal sampling index and the vertical sampling index of each sampling distance, physical sampling is performed on the exterior wall of the building to be detected to generate a multi-distance sampling image.

[0019] In the above scheme, the physical sampling of the exterior wall of the building to be detected based on the horizontal sampling index and the vertical sampling index of each sampling distance to generate a multi-distance sampling image includes:

[0020] Based on the horizontal sampling index of each sampling distance, the vertical sampling index of each sampling distance, the effective rectangular region resolution and the target resolution of each sampling distance, the physical sampling position of each sampling distance is obtained;

[0021] Based on the physical sampling location of each sampling distance, determine the sampled valid rectangular region and the unsampled valid rectangular region for each sampling distance;

[0022] Based on the sampled valid rectangular area and the unsampled valid rectangular area of ​​each sampling distance, thermal imaging is performed on the exterior wall of the building to be detected to generate the multi-distance sampling image.

[0023] In the above scheme, the step of performing thermal imaging on the exterior wall of the building to be detected based on the sampled valid rectangular area and the unsampled valid rectangular area at each sampling distance to generate the multi-distance sampling image includes:

[0024] For each sampling distance, calculate the average temperature of the sampled valid rectangular region, and based on the average temperature, determine the downsampled pixels of the sampled valid rectangular region;

[0025] Based on the average temperature of the sampled valid rectangular region, the average temperature of the unsampled valid rectangular region is calculated using a bilinear interpolation algorithm;

[0026] Based on the pixel values ​​of the four vertices of the sampled valid rectangular region, the initial pixel values ​​of the unsampled valid rectangular region are calculated using a bicubic interpolation algorithm.

[0027] The mean value of the unsampled valid rectangular region is mixed with the initial pixel value within the unsampled valid rectangular region to obtain the final pixel value of the unsampled valid rectangular region.

[0028] The pixel values ​​are combined based on the downsampled pixels of the sampled valid rectangular region and the final pixel values ​​of the unsampled valid rectangular region to obtain the sampled image for each sampling distance, thereby determining the multi-distance sampled image.

[0029] In the above scheme, the step of performing image fusion based on the multi-distance sampled images to generate wall temperature field images of the building exterior wall at different time points includes:

[0030] Image fusion is performed on the multi-distance sampled images using pre-determined distance weights to generate wall temperature field images of the building exterior wall at different time points; wherein, the distance weights represent the distance between the pixel location and the sampling point; or,

[0031] Based on the multi-distance sampling images, determine the maximum value of the corresponding pixels in the multi-distance sampling images; based on the maximum value, generate wall temperature field images of the exterior wall of the building to be detected at different time points; or...

[0032] The signal-to-noise ratio (SNR) of each of the multi-distance sampled images is calculated, and the fusion weight is determined based on the SNR. Image fusion is then performed based on the fusion weight to generate wall temperature field images of the building exterior wall at different time points.

[0033] In the above scheme, the step of performing external wall risk detection and risk level classification based on the temperature field image sequence to determine the current risk level of the external wall of the building to be detected includes:

[0034] The temperature field image sequence is normalized to obtain a normalized temperature field image sequence.

[0035] Using a pre-determined risk assessment model for wall detachment, the normalized temperature field image sequence is used to detect external wall risks, resulting in a probability map of wall detachment risk.

[0036] A risk assessment is performed based on the wall detachment risk probability map to obtain a comprehensive risk level; and a risk level classification is performed based on the comprehensive risk level and a preset risk threshold to determine the current risk level of the exterior wall of the building to be inspected.

[0037] Secondly, embodiments of this application provide a risk prediction system for external wall hollowing and detachment based on infrared detection. The system includes: an acquisition module, a generation module, and a determination module.

[0038] The acquisition module is used to acquire the wall material of the building exterior wall to be detected at different time points based on a preset time interval; and for the wall material at each of the different time points, acquire multiple sampling distances corresponding to the wall material from a preset mapping relationship library; wherein, the mapping relationship library is a mapping relationship between the wall material and multiple sampling distances;

[0039] The generation module is used to perform physical sampling on the exterior wall of the building to be detected based on the multiple sampling distances to generate multi-distance sampling images; and to perform image fusion based on the multi-distance sampling images to generate wall temperature field images of the exterior wall of the building to be detected at different time points.

[0040] The determining module is used to determine a temperature field image sequence based on the wall temperature field images at different time points; and to perform external wall risk detection and risk level classification based on the temperature field image sequence to determine the current risk level of the external wall of the building to be detected.

[0041] Thirdly, embodiments of this application provide a device for predicting the risk of hollow and detached exterior walls based on infrared detection, comprising: a processor and a memory; wherein,

[0042] The memory is used to store computer programs;

[0043] The processor is configured to call and run the computer program from the memory to perform the method as described in the first aspect.

[0044] Fourthly, embodiments of this application provide a computer-readable storage medium storing executable instructions for causing a processor to perform the method described in the first aspect.

[0045] This application provides a method and system for predicting the risk of hollowing and falling off of exterior walls based on infrared detection. The method includes: acquiring the wall material of the exterior wall of a building to be inspected at different time points based on a preset time interval; and for each wall material at different time points, acquiring multiple sampling distances corresponding to the wall material from a preset mapping relationship library; wherein, the mapping relationship library is a mapping relationship between wall materials and multiple sampling distances; performing physical sampling on the exterior wall of the building to be inspected based on the multiple sampling distances to generate multi-distance sampling images; performing image fusion based on the multi-distance sampling images to generate wall temperature field images of the exterior wall of the building to be inspected at different time points; and determining a temperature field image sequence based on the wall temperature field images at different time points; and performing exterior wall risk detection and risk level classification based on the temperature field image sequence to determine the current risk level of the exterior wall of the building to be inspected. In the above scheme, the multi-distance adaptive sampling method based on wall material effectively overcomes the problems of detail loss or field of view limitations that may exist in single-distance sampling by fusing thermal imaging information at multiple different spatial resolution scales. This generates a high-resolution wall temperature field image sequence rich in detail and accurate in temperature, laying a reliable data foundation for subsequent precise analysis. Based on the temperature field image sequence, external wall risk detection and risk level classification are performed to determine the current risk level of the building's external wall. The final results, presented as a risk probability map and risk level, transform safety management from experience-based judgment to data-driven approaches, significantly improving the scientific rigor and relevance of maintenance decisions. Furthermore, the rich detail and accurate temperature of the temperature field image sequence improve the accuracy and reliability of predicting external wall hollowing and detachment risks, reducing the rates of missed detections and false judgments. Attached Figure Description

[0046] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the specification, serve to explain the technical solutions of this application. Obviously, the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0047] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0048] Figure 1 A schematic diagram of an optional process for predicting the risk of hollowing and falling off of exterior walls based on infrared detection, provided in an embodiment of this application;

[0049] Figure 2 A schematic diagram of an optional sampling distance for a method for predicting the risk of hollowing and falling off of exterior walls based on infrared detection, provided in an embodiment of this application;

[0050] Figure 3 A schematic diagram of an optional effective rectangular area for a method for predicting the risk of hollowing and falling off of exterior walls based on infrared detection, provided in an embodiment of this application;

[0051] Figure 4 A schematic diagram of a risk prediction system for external wall hollowing and falling off based on infrared detection provided in this application embodiment;

[0052] Figure 5 This is a schematic diagram of a risk prediction device for hollow and falling exterior walls based on infrared detection, provided in an embodiment of this application. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the specific technical solutions of this application will be further described in detail below with reference to the accompanying drawings of the embodiments of this application. The following embodiments are used to illustrate this application, but are not intended to limit the scope of this application.

[0054] Unless otherwise defined, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. The terminology used in this application is for the purpose of describing embodiments of this application only and is not intended to be limiting of this application.

[0055] In the following description, references to "some embodiments," "this embodiment," "this application embodiment," and examples, etc., describe a subset of all possible embodiments. However, it is understood that "some embodiments" may be the same subset or different subset of all possible embodiments and may be combined with each other without conflict.

[0056] If the application documents contain similar descriptions such as "first / second", the following explanation shall be added: In the following description, the terms "first / second / third" are used only to distinguish similar objects and do not represent a specific order of objects. It is understood that "first / second / third" may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0057] This application provides a method for predicting the risk of exterior wall hollowing and detachment based on infrared detection. Figure 1 This is an optional flowchart illustrating an infrared detection-based method for predicting the risk of hollow and detached exterior walls, provided as an embodiment of this application. Figure 1 The steps shown are explained.

[0058] S101. Based on a preset time interval, obtain the wall material of the building exterior wall to be detected at different time points; and for the wall material at each time point in different time points, obtain multiple sampling distances corresponding to the wall material from a preset mapping relationship library; wherein, the mapping relationship library is the mapping relationship between the wall material and multiple sampling distances.

[0059] In some embodiments of this application, the preset time interval is a pre-set sampling time interval. The mapping relationship library contains the mapping relationship between wall materials and multiple sampling distances, and different wall materials correspond to different multiple sampling distances.

[0060] In some embodiments of this application, an infrared detection-based method for predicting the risk of hollow and detached exterior walls is adapted to the scenario of detecting the risk of hollow and detached exterior walls.

[0061] In some embodiments of this application, a method for predicting the risk of hollow and detached exterior walls based on infrared detection is adapted to an infrared detection-based system for predicting the risk of hollow and detached exterior walls.

[0062] In some embodiments of this application, the wall material of the building exterior wall to be detected at different time points is obtained based on a preset time interval; and for the wall material at each time point in different time points, multiple sampling distances corresponding to the wall material are obtained from a preset mapping relationship library.

[0063] For example, let's illustrate this using three sampling distances instead of multiple sampling distances. For common building exterior wall materials, including but not limited to ceramic tiles, stone, paint, and curtain wall glass, based on their thermophysical properties (such as emissivity and thermal conductivity) and the characteristics of hollow defects in thermal imaging, three sampling distances—near distance d1, medium distance d2, and far distance d3—are preset to form a mapping relationship library. The sampling distance is the straight-line distance from the center of the thermal imager lens to the surface of the wall being measured. The three specific sampling distance values ​​corresponding to this material are selected for on-site identification of the building exterior wall under test or to determine its main wall material based on architectural design data. According to the established mapping relationship library, the three specific sampling distance values ​​d1, d2, and d3 corresponding to this material are automatically selected. The three specific sampling distance values ​​are as follows: Figure 2 As shown, the three sampling distances can be extended to multiple sampling distances.

[0064] S102. Based on multiple sampling distances, perform physical sampling on the exterior wall of the building to be inspected to generate multi-distance sampling images.

[0065] In some embodiments of this application, the physical dimensions and target resolution of the exterior wall of the building to be detected are determined; and the image pixel size is determined based on the physical dimensions and target resolution; the effective rectangular region resolution corresponding to each of the multiple sampling distances is determined, as well as the sampling interval corresponding to each of the multiple sampling distances; wherein, the sampling interval includes a horizontal interval and a vertical interval; and the exterior wall of the building to be detected is physically sampled based on the multiple sampling distances, the image pixel size, the effective rectangular region resolution, and the sampling interval to generate a multi-distance sampling image.

[0066] S103. Based on multi-distance sampling images, perform image fusion to generate wall temperature field images of the building exterior wall at different time points; and determine the temperature field image sequence based on the wall temperature field images at different time points.

[0067] In some embodiments of this application, image fusion is performed on multiple distance-sampled images using predetermined distance weights to generate wall temperature field images of the building exterior wall at different time points. The distance weights represent the distance between a pixel location and a sampling point. Alternatively, based on the multiple distance-sampled images, the maximum value of corresponding pixels in the multiple distance-sampled images is determined; based on the maximum value, wall temperature field images of the building exterior wall at different time points are generated. Alternatively, the signal-to-noise ratio (SNR) of each of the multiple distance-sampled images is calculated, and a fusion weight is determined based on the SNR. Image fusion is then performed based on the fusion weights to generate wall temperature field images of the building exterior wall at different time points. A temperature field image sequence is determined based on the wall temperature field images at different time points.

[0068] S104. Based on the temperature field image sequence, perform external wall risk detection and risk level classification to determine the current risk level of the external wall of the building to be detected.

[0069] In some embodiments of this application, the temperature field image sequence is normalized to obtain a normalized temperature field image sequence; the normalized temperature field image sequence is then subjected to external wall risk detection using a pre-determined detachment risk assessment model to obtain a wall detachment risk probability map; a risk assessment is performed based on the wall detachment risk probability map to obtain a comprehensive risk level; and a risk level classification is performed based on the comprehensive risk level and a preset risk threshold to determine the current risk level of the building exterior wall to be inspected.

[0070] For example, addressing the shortcomings of existing infrared detection technologies for exterior walls, such as limited sampling methods and a lack of predictive power in analysis results, this application proposes a novel overall technical concept of "multi-scale perception-temporal prediction." The core lies in designing a complete technology chain from intelligent data acquisition to dynamic risk assessment. At the data acquisition end, a three-distance adaptive sampling mechanism based on wall material is employed. Through collaborative matrix-style acquisition at near, medium, and long distances, high-resolution details, mesoscale features, and overall thermal distribution information are obtained, respectively. A two-level fusion algorithm, employing "regional mean interpolation first, followed by pixel-level reconstruction," reconstructs the sparsely sampled data into a high-precision, fully covered wall temperature field, resolving the conflict between resolution and efficiency. At the data analysis end, a risk prediction model based on a spatiotemporal convolutional neural network is constructed. The system uses multiple high-precision temperature fields as spatiotemporal sequence inputs and automatically learns the evolution patterns of hollow hot spots in terms of morphology, area, and temperature gradient. The final output is a quantitative "risk probability map", which, combined with the dual-dimensional indicators of "current state severity" and "evolution trend speed", forms a scientific and intuitive risk level system, realizing the leap from hazard identification to risk warning.

[0071] Understandably, the multi-distance adaptive sampling method based on wall material effectively overcomes the potential for detail loss or limited field of view issues associated with single-distance sampling by fusing thermal imaging information from multiple different spatial resolution scales. This generates a high-resolution wall temperature field image sequence rich in detail and accurate in temperature, laying a reliable data foundation for subsequent precise analysis. Based on the temperature field image sequence, external wall risk detection and risk level classification are performed to determine the current risk level of the building's external wall. The final results, presented as a risk probability map and risk level, transform safety management from experience-based judgment to data-driven approaches, significantly improving the scientific rigor and relevance of maintenance decisions. Furthermore, the rich detail and accurate temperature of the temperature field image sequence enhance the accuracy and reliability of predicting external wall hollowing and detachment risks, reducing missed detections and false positives.

[0072] In some embodiments of this application, S102 can be implemented by S201-S203, as follows:

[0073] S201. Determine the physical dimensions and target resolution of the exterior wall of the building to be inspected; and determine the image pixel size based on the physical dimensions and target resolution.

[0074] For example, set the target resolution for the entire rectangular wall surface (i.e., the exterior wall of the building to be inspected):

[0075] Determine the actual physical width Δw and height Δh (i.e., the target resolution) for each pixel. Based on the actual physical width W and height H of the wall, calculate the image pixel size of the wall at the target resolution:

[0076] Image width in pixels: N_w = W / Δw;

[0077] Image height in pixels: N_h = H / Δh.

[0078] S202. Determine the effective rectangular region resolution corresponding to each of the multiple sampling distances, and the sampling interval corresponding to each of the multiple sampling distances; wherein, the sampling interval includes the horizontal interval and the vertical interval.

[0079] For example, the effective rectangular region is defined as follows: For each sampling distance di (i.e., d1, d2, d3), the physical range covered by a single thermal image is calculated based on the field of view and distance of the thermal imaging device, and an effective rectangular region is defined within this physical range. The corresponding effective rectangular region resolution (width v_w and height v_h) is obtained based on Δw and Δh.

[0080] A pre-defined sampling interval lookup table is used, which defines the physical distance sampling intervals corresponding to different wall sampling distances:

[0081] The sampling intervals corresponding to the close distance d1 are: horizontal interval h1 and vertical interval v1.

[0082] The sampling intervals corresponding to the mid-distance d2 are: horizontal interval h2 and vertical interval v2.

[0083] The sampling intervals corresponding to the long distance d3 are: horizontal interval h3 and vertical interval v3.

[0084] S203. Based on multiple sampling distances, image pixel size, effective rectangular area resolution, and sampling interval, physical sampling is performed on the exterior wall of the building to be detected to generate multi-distance sampling images.

[0085] In some embodiments of this application, the number of effective rectangular regions corresponding to each of the multiple sampling distances is calculated based on the image pixel size and the resolution of the effective rectangular region. The number of effective rectangular regions includes the number of horizontal effective rectangles and the number of vertical effective rectangles. For each of the multiple sampling distances, the horizontal sampling index and the vertical sampling index of each sampling distance are calculated based on the number of horizontal effective rectangles, the number of vertical effective rectangles, the horizontal interval, and the vertical interval. Based on the horizontal sampling index and the vertical sampling index of each sampling distance, physical sampling is performed on the exterior wall of the building to be detected to generate a multi-distance sampling image.

[0086] In some embodiments of this application, the physical sampling position of each sampling distance is obtained based on the horizontal sampling index of each sampling distance, the vertical sampling index of each sampling distance, the effective rectangular region resolution of each sampling distance, and the target resolution. Based on the physical sampling position of each sampling distance, the sampled effective rectangular region and the unsampled effective rectangular region of each sampling distance are determined. Based on the sampled effective rectangular region and the unsampled effective rectangular region of each sampling distance, thermal imaging is performed on the exterior wall of the building to be inspected to generate a multi-distance sampling image.

[0087] For example, the number of effective areas is calculated based on the target resolution of the rectangular wall and the resolution of the effective rectangular area, using the following formula:

[0088] Number of valid horizontal rectangular regions: num_w = N_w / v_w;

[0089] Number of valid vertical rectangular regions: num_h = N_h / v_h.

[0090] For any sampling distance, based on the number of horizontal and vertical effective rectangular regions, the horizontal and vertical intervals, and starting from the top left corner, the horizontal and vertical indices of the sampling area are calculated, as well as the number of horizontal and vertical sampling regions. Based on the horizontal and vertical indices of the sampling area, the resolution of the effective rectangular region, and the actual physical width Δw and height Δh, all physical sampling positions are obtained. Thermal imaging is then performed based on these physical sampling positions, and the images of the effective rectangular regions and the regions corresponding to the unsampled positions are combined to form images corresponding to any distance, namely T1, T2, and T3.

[0091] In some embodiments of this application, for each sampling distance, the average temperature of the sampled valid rectangular region is calculated, and based on the average temperature, the downsampled pixels of the sampled valid rectangular region are determined; based on the average temperature of the sampled valid rectangular region, the average value of the unsampled valid rectangular region is calculated using a bilinear interpolation algorithm; based on the pixel values ​​of the four vertices of the sampled valid rectangular region, the initial pixel values ​​of the unsampled valid rectangular region are calculated using a bicubic interpolation algorithm; the average value of the unsampled valid rectangular region is mixed with the initial pixel values ​​of the unsampled valid rectangular region to obtain the final pixel values ​​of the unsampled valid rectangular region; the pixel values ​​are combined based on the downsampled pixels of the sampled valid rectangular region and the final pixel values ​​of the unsampled valid rectangular region to obtain the sampling image for each sampling distance, thereby determining the multi-distance sampling image.

[0092] For example, the mean of the sampling area is calculated and interpolated: for each sampling distance dᵢ, the mean temperature of the sampled valid rectangular area is calculated and used as the known downsampled pixel; based on the mean temperature of all sampled valid rectangular areas, the mean of the unsampled valid rectangular area is calculated using a bilinear interpolation algorithm.

[0093] Generate initial pixel values ​​for the unsampled region: Based on the pixel values ​​of the four vertices of the sampled valid rectangular region, use a bicubic interpolation algorithm to calculate the initial pixel values ​​of all pixels within the unsampled valid rectangular region.

[0094] Generate single-distance thermal imaging images:

[0095] 1. For the unsampled valid rectangular region, its mean (from S31) is mixed with the initial pixel value (from S32) to obtain the final pixel value of the region;

[0096] 2. Combine the pixel values ​​of all valid rectangular areas (including sampled and unsampled areas) to generate a thermal imaging image at that sampling distance;

[0097] 3. Perform the above steps for each of the three sampling distances to obtain a near-range thermal imaging image T1, a mid-range thermal imaging image T2, and a long-range thermal imaging image T3. The near-range thermal imaging image T1, the mid-range thermal imaging image T2, and the long-range thermal imaging image T3 together determine the multi-distance sampling image.

[0098] Multi-distance sampling example: Building exterior wall inspection

[0099] 1. Basic information about the wall surface.

[0100] Suppose an office building's exterior wall needs to be inspected:

[0101] Wall physical dimensions: Width: W=20m; Height: H=30m

[0102] Target resolution: The actual physical size corresponding to each pixel is: Δw = 0.01m (1 cm) Δh = 0.01m (1 cm)

[0103] 2. Target resolution calculation.

[0104] Calculate the pixel size of the wall in the digital image based on the target resolution:

[0105] Image width in pixels: N_w = 20 / 0.01 = 2000 pixels;

[0106] Image height in pixels: N_h = 30 / 0.01 = 3000 pixels.

[0107] 3. Determine the sampling distance and the effective rectangular area.

[0108] Query the preset mapping database based on wall material (tile):

[0109] With a close distance d1 = 2m, the effective rectangular area resolution is 231*146.

[0110] Mid-range distance d2 = 5m, effective rectangular area resolution 577*364;

[0111] At a distance of d3=10m, the effective rectangular area resolution is 1122*728.

[0112] 4. Calculate the number of valid regions.

[0113] The nearest distance d1 = 2m; num_w = 2000 / 231 = 9; num_h = 3000 / 146 = 21 (rounded up);

[0114] Mid-distance d2 = 5m; num_w = 2000 / 577 = 4; num_h = 3000 / 364 = 9;

[0115] The distance d3 = 10m; num_w = 2000 / 1122 = 2; num_h = 3000 / 728 = 5.

[0116] 5. Obtain the sampling interval;

[0117] The horizontal interval h1 = 2 (sampling once every 2 valid regions), and the vertical interval v1 = 2;

[0118] Horizontal interval h2 = 3, vertical interval v2 = 4;

[0119] The horizontal interval h3 = 3, and the vertical interval v3 = 4.

[0120] 6. Example of physical sampling calculation (taking close range as an example).

[0121] like Figure 3 As shown, the sampling area is calculated:

[0122] Vertical sampling indices: 0, 3, 6 (3 in total)

[0123] Horizontal sampling index: 0, 5, 10, 15, 20 (5 in total)

[0124] Total number of sampling areas: 3 × 5 = 15 valid rectangular areas.

[0125] In some embodiments of this application, the image fusion based on multi-distance sampled images in S103 to generate wall temperature field images of the building exterior wall at different time points can be implemented through S301, S302, or S303, as follows:

[0126] S301. By using a predetermined distance weight, perform image fusion on the multi-distance sampled images to generate wall temperature field images of the building exterior wall at different time points; wherein, the distance weight represents the distance between the pixel position and the sampling point.

[0127] S302. Based on the multi-distance sampling images, determine the maximum value of the corresponding pixels in the multi-distance sampling images; based on the maximum value, generate wall temperature field images of the exterior wall of the building to be detected at different time points.

[0128] S303. Calculate the signal-to-noise ratio of each of the multi-distance sampled images, determine the fusion weight based on the signal-to-noise ratio, perform image fusion based on the fusion weight, and generate wall temperature field images of the building exterior wall at different time points.

[0129] For example, three thermal imaging images T1, T2, and T3 at different distances are aligned and registered;

[0130] The final wall temperature field image T_high_res is generated using a multi-scale fusion strategy. The fusion methods include:

[0131] a) Distance-based weighted fusion: Weights are assigned based on the distance between the pixel location and the sampling point, with higher weights for images closer to the pixel; i and j are pixel indices, and W1, W2, W3 are the distance weights corresponding to T1, T2, and T3.

[0132]

[0133] b) Maximum value fusion: Take the maximum value of corresponding pixels in the three images as the final value;

[0134]

[0135] c) Fusion based on quality assessment: The fusion weights are dynamically adjusted according to the image quality indicators such as signal-to-noise ratio and sharpness to obtain the adjusted fusion weights. Image fusion is performed based on the fusion weights to generate wall temperature field images of the building exterior wall at different time points. The specific fusion process is the same as a).

[0136] In some embodiments of this application, S104 can be implemented by S401-S403, as follows:

[0137] S401. Normalize the temperature field image sequence to obtain the normalized temperature field image sequence.

[0138] S402. Using a pre-determined risk assessment model for detachment, the normalized temperature field image sequence is used to detect the risk of detachment of the exterior wall, and a probability map of wall detachment risk is obtained.

[0139] In some embodiments of this application, a wall detachment risk probability map is obtained by extracting spatiotemporal features, performing spatiotemporal attention weighting, and reconstructing features on the normalized temperature field image sequence through a pre-determined detachment risk assessment model.

[0140] For example, construct a spatiotemporal convolutional neural network model.

[0141] Network architecture design

[0142] The spatiotemporal convolutional neural network employing an encoder-decoder architecture has the following specific structure in sequence:

[0143] Input layer:

[0144] Receive a normalized temperature field image sequence at multiple time points, with an input size of (T, H, W, 1).

[0145] T: Number of time steps (e.g., 4 quarters)

[0146] H, W: Image height and width

[0147] 1: Single-channel temperature data

[0148] Encoder section (spatiotemporal feature extraction):

[0149] 3D convolutional block 1:

[0150] 3D convolutional layer: kernel (3,3,3), number of channels 64, stride (1,1,1), padding same;

[0151] Batch Normalization Layer

[0152] ReLU activation function

[0153] 3D max pooling layer: pooling window (2,2,2), stride (2,2,2);

[0154] 3D Convolutional Block 2:

[0155] 3D convolutional layer: kernel (3,3,3), number of channels 128, stride (1,1,1), padding same;

[0156] Batch Normalization Layer

[0157] ReLU activation function

[0158] 3D max pooling layer: pooling window (2,2,2), stride (2,2,2);

[0159] 3D Convolutional Block 3:

[0160] 3D convolutional layer: kernel (3,3,3), number of channels 256, stride (1,1,1), padding same;

[0161] Batch Normalization Layer

[0162] ReLU activation function

[0163] 3D max pooling layer: pooling window (2,2,2), stride (2,2,2);

[0164] Spatiotemporal attention module:

[0165] Time attention: Applying a self-attention mechanism to the time dimension to learn the importance weights of different time points;

[0166] Spatial attention: Apply attention to the spatial location of the feature map, focusing on areas of significant hot spot changes;

[0167] Channel attention: The SE module learns the importance of different feature channels;

[0168] Decoder section (feature reconstruction):

[0169] 3D deconvolution block 1:

[0170] 3D deconvolution layer: kernel (3,3,3), number of channels 128, stride (2,2,2);

[0171] Concatenate with the corresponding layer features of the encoder (skip connection);

[0172] 3D convolutional layer: kernel (3,3,3), number of channels 128;

[0173] ReLU activation function

[0174] 3D deconvolution block 2:

[0175] 3D deconvolution layer: kernel (3,3,3), number of channels 64, stride (2,2,2);

[0176] Concatenate with the corresponding layer features of the encoder.

[0177] 3D convolutional layer: kernel (3,3,3), number of channels 64;

[0178] ReLU activation function

[0179] 3D deconvolution block 3:

[0180] 3D deconvolution layer: kernel (3,3,3), number of channels 32, stride (2,2,2);

[0181] 3D convolutional layer: kernel (3,3,3), number of channels 32;

[0182] ReLU activation function

[0183] Output layer:

[0184] 3D convolutional layer: kernel (1,1,1), number of channels 1;

[0185] Sigmoid activation function

[0186] Output shape: (1, H, W, 1), representing the risk probability map P_risk;

[0187] During training,

[0188] 1. The loss function design adopts a weighted composite loss function:

[0189] L_total = α·L_wbce + β·L_dice

[0190] Weighted binary cross-entropy loss (L_wbce): handles imbalance between positive and negative samples, assigning higher weights to high-risk pixels;

[0191] Dice loss (L_dice): Improves the detection performance of small hot spots;

[0192] Weight settings: α=0.7, β=0.3.

[0193] 2. Training Strategies

[0194] Three-stage training method:

[0195] Phase 1: Pre-trained encoder-decoder (learning rate 1e-3, 100 rounds);

[0196] Phase 2: Unfreezing and fine-tuning the spatiotemporal attention module (learning rate 1e-4, 50 rounds);

[0197] Phase 3: Overall network fine-tuning (learning rate 1e-5, 20 rounds).

[0198] Key training configuration:

[0199] Optimizer: Adam (β1=0.9, β2=0.999);

[0200] Batch size: 8-16 (adjust according to GPU memory);

[0201] Learning rate scheduling: cosine annealing + early stopping mechanism;

[0202] Regularization: Dropout(0.3) + L2 weight decay(1e-4).

[0203] S403. Conduct a risk assessment based on the wall detachment risk probability map to obtain a comprehensive risk level; and classify the risk level based on the comprehensive risk level and the preset risk threshold to determine the current risk level of the exterior wall of the building to be inspected.

[0204] An example of a risk assessment indicator extraction method:

[0205] 1. Extraction of high-risk area ratio: Threshold segmentation is performed on the risk probability map P_risk, the number of pixels with a risk probability greater than 0.7 is counted, and the percentage of them in the total number of pixels is calculated.

[0206] 2. Calculation of average risk probability: The average risk probability value of the wall is obtained by averaging the pixel values ​​of all pixels in the risk probability map P_risk.

[0207] 3. Extraction of the length of continuous high-risk regions: High-risk pixel regions are identified using a connected component analysis algorithm, and the physical length of the largest connected region is calculated.

[0208] 4. Risk growth rate calculation: Compare the current average risk probability with the previous test, and calculate the risk probability growth rate per unit time.

[0209] 5. Overall risk level = High-risk area ratio weight × High-risk area ratio + Average risk weight × Average risk probability + Continuous length weight × Standardized continuous length + Growth rate weight × Standardized growth rate.

[0210] Risk level classification:

[0211] Four threshold levels are set based on the overall risk level:

[0212] Level I (Safe): Overall risk level is below the first threshold;

[0213] Level II (Attention): The overall risk level is within the second threshold range;

[0214] Level III (Warning): The overall risk level is within the third threshold range;

[0215] Level IV (Hazardous): The overall risk level is higher than the third threshold.

[0216] Method for determining the overall risk level of a wall structure:

[0217] Determine the current overall risk level and risk classification criteria, and determine the current risk level.

[0218] The beneficial effects of this application are as follows:

[0219] 1. Improved the accuracy and reliability of detection, and reduced the rate of missed detections and false judgments.

[0220] This application employs a three-level adaptive sampling method based on wall material: near-field, mid-field, and long-field. By fusing thermal imaging information from three different spatial resolution scales and using a unique two-stage interpolation reconstruction algorithm (first interpolating the region mean, then interpolating the pixel values), it effectively overcomes the potential for detail loss or limited field of view issues in single-distance sampling. This generates a high-resolution wall temperature field with rich detail and accurate temperature, laying a reliable data foundation for subsequent precise analysis.

[0221] 2. It provides quantitative and visualized risk assessment results to support scientific decision-making.

[0222] This application not only outputs qualitative hotspot images, but also generates a "risk probability map" for wall detachment using a neural network model. The final result, with the risk probability map and risk level, transforms safety management from experience-based judgment to data-driven approaches, significantly improving the scientific rigor and relevance of maintenance decisions.

[0223] Based on the above embodiments, this application also provides an infrared detection-based method for predicting the risk of exterior wall hollowing and detachment. Figure 4 As shown, Figure 4 This application provides a schematic diagram of a risk prediction system for external wall hollowing and detachment based on infrared detection. The risk prediction system 4 for external wall hollowing and detachment based on infrared detection includes: an acquisition module 401, a generation module 402, and a determination module 403, wherein...

[0224] The acquisition module 401 is used to acquire the wall material of the building exterior wall to be detected at different time points based on a preset time interval; and for the wall material at each of the different time points, acquire multiple sampling distances corresponding to the wall material from a preset mapping relationship library; wherein, the mapping relationship library is a mapping relationship between the wall material and multiple sampling distances;

[0225] The generation module 402 is used to perform physical sampling on the exterior wall of the building to be detected based on the multiple sampling distances to generate multi-distance sampling images; and to perform image fusion based on the multi-distance sampling images to generate wall temperature field images of the exterior wall of the building to be detected at different time points.

[0226] The determining module 403 is used to determine a temperature field image sequence based on the wall temperature field images at different time points; and to perform external wall risk detection and risk level classification based on the temperature field image sequence to determine the current risk level of the external wall of the building to be detected.

[0227] In some embodiments of this application, the determining module 403 is further configured to determine the physical dimensions and target resolution of the exterior wall of the building to be detected; and based on the physical dimensions and target resolution, determine the image pixel size; determine the effective rectangular region resolution corresponding to each of the plurality of sampling distances, and the sampling interval corresponding to each of the plurality of sampling distances; wherein the sampling interval includes a horizontal interval and a vertical interval; and perform physical sampling on the exterior wall of the building to be detected based on the plurality of sampling distances, the image pixel size, the effective rectangular region resolution, and the sampling interval to generate a multi-distance sampling image.

[0228] In some embodiments of this application, the generation module 402 is further configured to calculate the number of effective rectangular regions corresponding to each of the plurality of sampling distances based on the image pixel size and the effective rectangular region resolution; wherein, the number of effective rectangular regions includes the number of horizontal effective rectangles and the number of vertical effective rectangles; for each of the plurality of sampling distances, calculate the horizontal sampling index and the vertical sampling index of each sampling distance according to the number of horizontal effective rectangles, the number of vertical effective rectangles, the horizontal interval, and the vertical interval; and perform physical sampling on the exterior wall of the building to be detected based on the horizontal sampling index and the vertical sampling index of each sampling distance to generate a multi-distance sampling image.

[0229] In some embodiments of this application, the generation module 402 is further configured to obtain the physical sampling position of each sampling distance based on the horizontal sampling index of each sampling distance, the vertical sampling index of each sampling distance, the effective rectangular region resolution of each sampling distance, and the target resolution; determine the sampled effective rectangular region and the unsampled effective rectangular region of each sampling distance based on the physical sampling position of each sampling distance; and perform thermal imaging on the exterior wall of the building to be detected based on the sampled effective rectangular region and the unsampled effective rectangular region of each sampling distance to generate the multi-distance sampling image.

[0230] In some embodiments of this application, the determining module 403 is further configured to: calculate the average temperature of the sampled valid rectangular region for each sampling distance; determine the downsampled pixels of the sampled valid rectangular region based on the average temperature; calculate the average value of the unsampled valid rectangular region based on the average temperature of the sampled valid rectangular region using a bilinear interpolation algorithm; calculate the initial pixel value of the unsampled valid rectangular region based on the four vertex pixel values ​​of the sampled valid rectangular region using a bicubic interpolation algorithm; mix the average value of the unsampled valid rectangular region with the initial pixel value of the unsampled valid rectangular region to obtain the final pixel value of the unsampled valid rectangular region; and combine the pixel values ​​based on the downsampled pixels of the sampled valid rectangular region and the final pixel value of the unsampled valid rectangular region to obtain the sampling image for each sampling distance, thereby determining the multi-distance sampling image.

[0231] In some embodiments of this application, the generation module 402 is further configured to perform image fusion on the multi-distance sampling images using a pre-determined distance weight to generate wall temperature field images of the building exterior wall to be detected at different time points; wherein, the distance weight represents the distance between the pixel position and the sampling point; or, based on the multi-distance sampling images, determine the maximum value of the corresponding pixel in the multi-distance sampling images; based on the maximum value, generate wall temperature field images of the building exterior wall to be detected at different time points; or, calculate the signal-to-noise ratio of each of the multi-distance sampling images, determine the fusion weight based on the signal-to-noise ratio, perform image fusion based on the fusion weight, and generate wall temperature field images of the building exterior wall to be detected at different time points.

[0232] In some embodiments of this application, the determining module 403 is further configured to normalize the temperature field image sequence to obtain a normalized temperature field image sequence; perform external wall risk detection on the normalized temperature field image sequence using a pre-determined detachment risk assessment model to obtain a wall detachment risk probability map; perform risk assessment based on the wall detachment risk probability map to obtain a comprehensive risk level; and classify the risk level based on the comprehensive risk level and a preset risk threshold to determine the current risk level of the building exterior wall to be detected.

[0233] Based on the above embodiments, this application also provides an infrared detection-based device for predicting the risk of exterior wall hollowing and detachment. Figure 5 As shown, Figure 5This is a schematic diagram of a structural device for predicting the risk of exterior wall hollowing and detachment based on infrared detection, provided in an embodiment of this application. The device 5 includes a processor 501 and a memory 502. The memory 502 is used to store computer programs; the processor 501 is used to call and run the computer programs from the memory to execute the method for predicting the risk of exterior wall hollowing and detachment based on infrared detection as described in the above embodiment.

[0234] In the embodiments of this application, the processor 501 described above can be at least one of the following: Application-Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), Central Processing Unit (CPU), Controller, Microcontroller, and Microprocessor. It is understood that for different devices, the electronic device used to implement the above processor function can also be other types, and the embodiments of this application do not specifically limit it.

[0235] This application provides a computer-readable storage medium storing a computer program for implementing, when executed by a processor, a method for predicting the risk of hollowing and falling off of exterior walls based on infrared detection, as described in any of the above embodiments.

[0236] For example, the program instructions corresponding to the infrared detection-based method for predicting the risk of hollow and falling off exterior walls in this embodiment can be stored on storage media such as optical discs, hard disks, and USB flash drives. When the program instructions corresponding to the infrared detection-based method for predicting the risk of hollow and falling off exterior walls in the storage media are read or executed by an electronic device, the infrared detection-based method for predicting the risk of hollow and falling off exterior walls as described in any of the above embodiments can be realized.

[0237] Furthermore, in the embodiments of this application, the functional modules can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional module.

[0238] If the integrated unit is implemented as a software functional module and is not sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this embodiment, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the method of this embodiment. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0239] It should be understood that the phrases "one embodiment," "an embodiment," or "some embodiments" mentioned throughout the specification mean that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "in one embodiment," "in one embodiment," or "in some embodiments" appearing throughout the specification do not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. It should be understood that in the various embodiments of this application, the sequence numbers of the above-described processes do not imply a sequential order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. The sequence numbers of the embodiments in this application are merely for descriptive purposes and do not represent the superiority or inferiority of the embodiments. The descriptions of the various embodiments above tend to emphasize the differences between the various embodiments; their similarities or commonalities can be referred to mutually, and for the sake of brevity, these will not be repeated here.

[0240] The modules described above as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules. They may be located in one place or distributed across multiple network units. Some or all of the modules may be selected to achieve the purpose of this embodiment according to actual needs.

[0241] In addition, each functional module in the various embodiments of this application can be integrated into one processing unit, or each module can be a separate unit, or two or more modules can be integrated into one unit; the integrated modules can be implemented in hardware or in the form of hardware plus software functional units.

[0242] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as mobile storage devices, read-only memory (ROM), magnetic disks, or optical disks.

[0243] The methods disclosed in the several method embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments.

[0244] The features disclosed in the several product embodiments provided in this application can be arbitrarily combined without conflict to obtain new product embodiments.

[0245] The features disclosed in the several method or device embodiments provided in this application can be arbitrarily combined without conflict to obtain new method or device embodiments.

[0246] The above description is merely an embodiment of this application, but the protection scope of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the protection scope of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.

Claims

1. A method for predicting the risk of hollow and detached exterior walls based on infrared detection, characterized in that, The method includes: Based on a preset time interval, the wall material of the building exterior wall to be detected at different time points is obtained; and for the wall material at each of the different time points, multiple sampling distances corresponding to the wall material are obtained from a preset mapping relationship library; wherein, the mapping relationship library is a mapping relationship between the wall material and multiple sampling distances; Based on the multiple sampling distances, physical sampling is performed on the exterior wall of the building to be detected to generate multi-distance sampling images; Based on the multi-distance sampled images, image fusion is performed to generate wall temperature field images of the building exterior wall at different time points; and based on the wall temperature field images at different time points, a temperature field image sequence is determined. Based on the temperature field image sequence, external wall risk detection and risk level classification are performed to determine the current risk level of the external wall of the building to be detected. The step of physically sampling the exterior wall of the building to be detected based on the multiple sampling distances to generate a multi-distance sampling image includes: Determine the physical dimensions and target resolution of the exterior wall of the building to be inspected; and determine the image pixel size based on the physical dimensions and target resolution. Determine the effective rectangular region resolution corresponding to each of the plurality of sampling distances, and the sampling interval corresponding to each of the plurality of sampling distances; wherein, the sampling interval includes a horizontal interval and a vertical interval; Based on the image pixel size and the effective rectangular region resolution, the number of effective rectangular regions corresponding to each of the multiple sampling distances is calculated; wherein, the number of effective rectangular regions includes the number of horizontal effective rectangles and the number of vertical effective rectangles; For each of the plurality of sampling distances, the horizontal sampling index and the vertical sampling index of each sampling distance are calculated based on the number of horizontal valid rectangles, the number of vertical valid rectangles, the horizontal interval, and the vertical interval. Based on the horizontal sampling index and the vertical sampling index of each sampling distance, physical sampling is performed on the exterior wall of the building to be detected to generate a multi-distance sampling image.

2. The method according to claim 1, characterized in that, The method of physically sampling the exterior wall of the building to be detected based on the horizontal sampling index and the vertical sampling index of each sampling distance to generate a multi-distance sampling image includes: Based on the horizontal sampling index of each sampling distance, the vertical sampling index of each sampling distance, the effective rectangular region resolution and the target resolution of each sampling distance, the physical sampling position of each sampling distance is obtained; Based on the physical sampling location of each sampling distance, determine the sampled valid rectangular region and the unsampled valid rectangular region for each sampling distance; Based on the sampled valid rectangular area and the unsampled valid rectangular area of ​​each sampling distance, thermal imaging is performed on the exterior wall of the building to be detected to generate the multi-distance sampling image.

3. The method according to claim 2, characterized in that, The process involves performing thermal imaging on the exterior wall of the building under test based on the sampled valid rectangular area and the unsampled valid rectangular area for each sampling distance, generating the multi-distance sampling image, including: For each sampling distance, calculate the average temperature of the sampled valid rectangular region, and based on the average temperature, determine the downsampled pixels of the sampled valid rectangular region; Based on the average temperature of the sampled valid rectangular region, the average temperature of the unsampled valid rectangular region is calculated using a bilinear interpolation algorithm; Based on the pixel values ​​of the four vertices of the sampled valid rectangular region, the initial pixel values ​​of the unsampled valid rectangular region are calculated using a bicubic interpolation algorithm. The mean value of the unsampled valid rectangular region is mixed with the initial pixel value within the unsampled valid rectangular region to obtain the final pixel value of the unsampled valid rectangular region. The pixel values ​​are combined based on the downsampled pixels of the sampled valid rectangular region and the final pixel values ​​of the unsampled valid rectangular region to obtain the sampled image for each sampling distance, thereby determining the multi-distance sampled image.

4. The method according to claim 1, characterized in that, The step of performing image fusion based on the multi-distance sampled images to generate wall temperature field images of the building exterior wall at different time points includes: Image fusion is performed on the multi-distance sampled images using pre-determined distance weights to generate wall temperature field images of the building exterior wall at different time points; wherein, the distance weights represent the distance between the pixel location and the sampling point; or, Based on the multi-distance sampling images, determine the maximum value of the corresponding pixels in the multi-distance sampling images; based on the maximum value, generate wall temperature field images of the exterior wall of the building to be detected at different time points; or... The signal-to-noise ratio (SNR) of each of the multi-distance sampled images is calculated, and the fusion weight is determined based on the SNR. Image fusion is then performed based on the fusion weight to generate wall temperature field images of the building exterior wall at different time points.

5. The method according to claim 1, characterized in that, The step of performing external wall risk detection and risk level classification based on the temperature field image sequence to determine the current risk level of the external wall of the building to be detected includes: The temperature field image sequence is normalized to obtain a normalized temperature field image sequence. Using a pre-determined risk assessment model for wall detachment, the normalized temperature field image sequence is used to detect external wall risks, resulting in a probability map of wall detachment risk. A risk assessment is performed based on the wall detachment risk probability map to obtain a comprehensive risk level; and a risk level classification is performed based on the comprehensive risk level and a preset risk threshold to determine the current risk level of the exterior wall of the building to be inspected.

6. A risk prediction system for hollow and detached exterior walls based on infrared detection, characterized in that, The infrared detection-based risk prediction system for external wall hollowing and detachment includes: an acquisition module, a generation module, and a determination module, wherein... The acquisition module is used to acquire the wall material of the building exterior wall to be detected at different time points based on a preset time interval; and for the wall material at each of the different time points, acquire multiple sampling distances corresponding to the wall material from a preset mapping relationship library; wherein, the mapping relationship library is a mapping relationship between the wall material and multiple sampling distances; The generation module is used to perform physical sampling on the exterior wall of the building to be detected based on the multiple sampling distances to generate multi-distance sampling images; and to perform image fusion based on the multi-distance sampling images to generate wall temperature field images of the exterior wall of the building to be detected at different time points. The determining module is used to determine a temperature field image sequence based on the wall temperature field images at different time points; and to perform external wall risk detection and risk level classification based on the temperature field image sequence to determine the current risk level of the external wall of the building to be detected. The generation module is further configured to: determine the physical dimensions and target resolution of the exterior wall of the building to be detected; determine the image pixel size based on the physical dimensions and target resolution; determine the effective rectangular region resolution corresponding to each of the plurality of sampling distances, and the sampling interval corresponding to each of the plurality of sampling distances; wherein the sampling interval includes a horizontal interval and a vertical interval; calculate the number of effective rectangular regions corresponding to each of the plurality of sampling distances based on the image pixel size and the effective rectangular region resolution; wherein the number of effective rectangular regions includes the number of horizontal effective rectangles and the number of vertical effective rectangles; for each of the plurality of sampling distances, calculate the horizontal sampling index and the vertical sampling index of each sampling distance based on the number of horizontal effective rectangles, the number of vertical effective rectangles, the horizontal interval, and the vertical interval; and perform physical sampling on the exterior wall of the building to be detected based on the horizontal sampling index and the vertical sampling index of each sampling distance to generate a multi-distance sampling image.

7. A device for predicting the risk of hollowing and falling off of exterior walls based on infrared detection, characterized in that, include: Processor and memory, of which, The memory is used to store computer programs; The processor is configured to call and run the computer program from the memory to perform the method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, It stores executable instructions for causing a processor to execute, thereby implementing the method of any one of claims 1 to 5.

Images