A thermal imaging picture temperature recognition calculation method based on unmanned aerial vehicle shooting

Abstract

The application discloses a kind of based on unmanned aerial vehicle shooting's thermal imaging chart temperature identification calculation method.It includes the following steps:S1, data acquisition: target area thermal imaging chart data and environmental temperature data are collected, and the time synchronization of acquisition data is carried out;S2, the thermal imaging image data collected is handled;S3, the mapping relationship model of thermal imaging image data and temperature data is established;S4, temperature prediction and correction are carried out;S5, to maintain the precision of model in long-term operation, dynamic updating mechanism is established.The application realizes the temperature identification and monitoring of wide range area non-contact, high precision, adaptive environmental change.

Description

Technical Field

[0001] This invention relates to the field of infrared thermal imaging, and in particular to a method for temperature recognition and calculation based on thermal images taken by drones. Background Technology

[0002] Accurate identification and monitoring of the temperature of objects or environments is of paramount importance in many fields. Traditional temperature measurement methods, such as contact thermometers and thermocouples, while meeting the requirements to some extent, have numerous limitations.

[0003] In industrial production, contact temperature measurement equipment requires direct contact with the object being measured. This presents significant challenges for high-temperature, high-pressure, high-speed, or corrosive materials, potentially damaging the equipment and impacting production safety and efficiency. For instance, in steel smelting, where molten steel reaches extremely high temperatures, contact measurement is difficult to implement and results in severe equipment wear and tear.

[0004] In the medical field, contact-based temperature measurement can cause discomfort to patients, especially infants, young children, and critically ill patients. Furthermore, during pandemic control efforts, rapid temperature screening of large numbers of people is necessary, and contact-based measurement is inefficient and carries the risk of cross-infection.

[0005] In the field of security monitoring, traditional temperature monitoring methods cannot achieve real-time, non-contact monitoring of large areas. For example, in forest fire early warning, it is necessary to detect potential fire sources in a timely manner, and traditional methods are difficult to achieve rapid and accurate location.

[0006] With the development of technology, thermal imaging technology has emerged. Thermal imaging devices can receive the infrared radiation emitted by objects and convert it into thermal images. However, currently, simple thermal images can only visually show the temperature distribution of an object, lacking the ability to accurately quantify and identify temperature.

[0007] The formation of thermal images is easily affected by environmental factors, such as ambient temperature, humidity, and particulate matter in the atmosphere. These factors can interfere with the infrared radiation emitted by objects, causing deviations in the temperature information of the thermal image and resulting in reduced accuracy of the final temperature recognition. For example, in high-temperature industrial environments, the heat emitted by surrounding equipment can affect the accurate measurement of the target object's temperature.

[0008] In addition, the formation of thermal images is also affected by emissivity and atmospheric attenuation. Since the emissivity of a material surface mainly depends on the material properties and surface condition, the emissivity needs to be measured by measuring the infrared radiation on the surface of the electrical equipment. That is, if the infrared diagnostic instrument receives the same infrared radiation power from the target, the results will vary depending on the emissivity of the target equipment surface.

[0009] When using a thermal imager to inspect equipment in an outdoor store, the instrument receives radiation emitted by the corresponding part of the equipment itself, as well as reflections from other backgrounds and direct solar radiation. All of this radiation will interfere with the temperature of the part of the equipment being tested, leading to errors in fault detection.

[0010] Limitations of the equipment itself: Different brands and models of thermal imaging equipment have differences in detector sensitivity, resolution, and other performance characteristics. Even the same equipment may experience performance degradation after prolonged use, leading to a decrease in the quality of thermal images and consequently affecting the accuracy of temperature identification.

[0011] The technical problem to be solved by this invention is as follows:

[0012] 1. Insufficient spatial coverage: Solving the problem that traditional methods cannot cover large areas;

[0013] 2. Limited recognition accuracy: Improve the accuracy of temperature recognition in complex environments;

[0014] 3. Real-time response delay: Enables real-time temperature calculation under dynamic environments;

[0015] 4. Poor model universality: Enhance the adaptability of the algorithm in different application scenarios. Summary of the Invention

[0016] To address the existing problems, this invention provides a method for temperature recognition and calculation based on thermal images captured by drones, the specific solution of which is as follows:

[0017] A method for temperature recognition and calculation based on thermal images taken by drones includes the following steps:

[0018] S1, Data Acquisition: Acquire thermal imaging data and ambient temperature data of the target area, and synchronize the acquired data in time;

[0019] S2, Process the acquired thermal imaging image data;

[0020] S3, Establish a mapping model between thermal imaging image data and temperature data;

[0021] S4 performs temperature prediction and correction;

[0022] S5. To maintain the accuracy of the model during long-term operation, a dynamic update mechanism is established.

[0023] Preferably, in step S1, a high-resolution thermal imaging camera mounted on a drone is used to collect thermal imaging data of the target area, and multiple temperature sensors deployed within the target area are used to collect ambient temperature data.

[0024] Preferably, the image data processing in step S2 includes:

[0025] S21. Image preprocessing: Filter the raw thermal image captured by the UAV to remove noise and perform histogram equalization to enhance image contrast.

[0026] S22. Target Region Segmentation: The Otsu adaptive threshold segmentation algorithm is used to binarize the preprocessed image and distinguish the foreground target of interest from the irrelevant background.

[0027] S23. Feature Extraction: For each segmented connected region of the foreground target, calculate its pixel grayscale statistical features, such as the average grayscale value or median value of the region, as pixel feature values ​​representing the temperature level of the region.

[0028] Preferably, step S3 includes the following steps:

[0029] S31, pairing the pixel feature values ​​of thermal imaging with the actual temperature;

[0030] S32 uses multiple linear regression to construct a mathematical model for the mapping: Where T is the actual temperature value, These are pixel feature values. ϵ is the regression coefficient, and ϵ is the error term;

[0031] S33, perform model training and validation, and use cross-validation to evaluate the model's accuracy.

[0032] Preferably, in step S4, the established model is used to predict the temperature of the newly acquired thermal image, and the mapping relationship model established in step S3 is updated periodically according to real-time environmental changes to maintain the accuracy of the prediction.

[0033] Preferably, the dynamic update mechanism includes the following steps:

[0034] S51, Data Input: Input the raw data into the model for processing;

[0035] S52, Model Evaluation: Used to judge the model's performance on the current data; evaluation metrics include accuracy, precision, recall, and F1 score. The evaluation process is carried out using cross-validation to ensure the model's generalization ability.

[0036] S53, Parameter Update: In order to improve the performance of the model and reduce errors, the parameters of the model are adjusted according to the results of the model evaluation.

[0037] S54, Model Deployment: Deploy the model, after evaluation and parameter updates, to the production environment for prediction of new data and output the prediction results;

[0038] S55, Performance Monitoring: Continuously compare the identified temperature value output by S4 with the reference temperature value measured by the sensor in the target area, and calculate the average absolute error.

[0039] S56, Error Feedback: When the mean absolute error exceeds a preset threshold, it is determined that the model performance has degraded, and an error feedback signal is generated;

[0040] S57, Parameter Update: Based on the error feedback signal, a new round of data acquisition and model training process is initiated. The temperature mapping relationship model in S3 is retrained using the new dataset to update the regression coefficients.

[0041] The present invention also discloses a computer-readable storage medium storing a computer program, which, when executed, performs the method described in any of the above-mentioned embodiments.

[0042] The present invention also discloses a computer system including a processor, a storage medium storing a computer program, and the processor reading from the storage medium and running the computer program to perform the method described in any of the preceding claims.

[0043] The beneficial effects of this invention are as follows:

[0044] 1. High precision and strong robustness: By integrating on-site measured data and environmental parameters through a multiple linear regression model, the effects of emissivity differences, environmental radiation interference and atmospheric attenuation are effectively compensated, significantly improving the temperature recognition accuracy in complex environments.

[0045] 2. Applicability to equipment and scenarios: This method does not rely on the absolute temperature calibration of thermal imaging equipment at the factory, but establishes a mapping relationship through field data learning. Therefore, it can be adapted to different brands and models of drone thermal imaging equipment and is applicable to various scenarios such as industry, agriculture, and security.

[0046] 3. Dynamic Adaptive Capability: The unique dynamic update mechanism can periodically correct model parameters, enabling the system to adapt to gradual changes in environmental factors such as seasons and weather over a long period of time, thus solving the problem of static model performance decaying over time.

[0047] 4. Large-scale real-time monitoring: Combined with a drone platform, it can achieve rapid, non-contact temperature inspection of a large area, and realize real-time or near real-time temperature identification and alarm through airborne or edge computing, with fast response speed.

[0048] 5. Automation and intelligence: From data collection, processing, modeling to updating, the entire process can be highly automated, reducing manual intervention and lowering operation and maintenance costs. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 This is a flowchart of the method of the present invention;

[0051] Figure 2 This is a flowchart of the dynamic update process in step S5 of the present invention;

[0052] Figure 3 This is a schematic diagram of image processing in step S2 of the present invention;

[0053] Figure 4 This is a schematic diagram illustrating the mapping relationship between thermal imaging image data and temperature data in this invention. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] like Figure 1 A method for temperature recognition and calculation based on thermal images taken by drones includes the following steps:

[0056] S1, Data Acquisition: Acquire thermal imaging data and ambient temperature data of the target area, and synchronize the acquired data in time;

[0057] Specifically, the drone configuration includes: equipping the drone with a high-resolution thermal imaging camera to ensure clear thermal imaging of the target area under different weather conditions.

[0058] On-site temperature measurement: Multiple high-precision temperature sensors are deployed in the monitoring area to record ambient temperature data in real time.

[0059] Data synchronization: Ensure that thermal imaging data and temperature sensor data are synchronized in time.

[0060] S2 processes the acquired thermal imaging image data.

[0061] Specifically, such as Figure 3 The image data processing in step S2 includes:

[0062] S21. Image preprocessing: Filter the raw thermal image captured by the UAV to remove noise and perform histogram equalization to enhance image contrast.

[0063] S22. Target Region Segmentation: The Otsu adaptive threshold segmentation algorithm is used to binarize the preprocessed image and distinguish the foreground target of interest from the irrelevant background.

[0064] S23. Feature Extraction: For each segmented connected region of the foreground target, calculate its pixel grayscale statistical features, such as the average grayscale value or median value of the region, as pixel feature values ​​representing the temperature level of the region.

[0065] S3. Establish a mapping model between thermal imaging image data and temperature data.

[0066] like Figure 4 Specifically, it includes the following steps:

[0067] S31, pairing the pixel feature values ​​of thermal imaging with the actual temperature;

[0068] S32 uses multiple linear regression to construct a mathematical model for the mapping: Where T is the actual temperature value, These are pixel feature values. ϵ is the regression coefficient, and ϵ is the error term;

[0069] S33, perform model training and validation, and use cross-validation to evaluate the model's accuracy.

[0070] S4 performs temperature prediction and correction;

[0071] Temperature prediction: Predict the temperature in new images using the established model.

[0072] Correction mechanism: The temperature mapping model is updated periodically in response to environmental changes.

[0073] Establish the temperature-time response function:

[0074]

[0075] Achieve dynamic environmental compensation.

[0076] S5. To maintain the accuracy of the model during long-term operation, a dynamic update mechanism is established.

[0077] like Figure 2 The dynamic update mechanism includes the following steps:

[0078] S51, Data Input: This is the starting point of the process, referring to the input of raw data into the model for processing. Data can include feature variables (input features) and target variables (labels or results). The quality and diversity of the data are crucial to the model's performance;

[0079] S52, Model Evaluation: After data input, the model evaluates the input data. The purpose of evaluation is to determine the model's performance on the current data. Commonly used evaluation metrics include accuracy, precision, recall, and F1 score. The evaluation process can be conducted using methods such as cross-validation to ensure the model's generalization ability.

[0080] S53, Parameter Update: In order to improve the performance of the model and reduce errors, the parameters of the model are adjusted based on the results of the model evaluation.

[0081] S54, Model Deployment: Once the model has been evaluated and its parameters updated, the next step is to deploy it to the production environment to make predictions on new data. After deployment, the model can process data in real time or in batches and output prediction results.

[0082] S55, Performance Monitoring: After model deployment, its performance must be continuously monitored. This includes tracking the model's performance on new data to ensure it maintains its expected accuracy and reliability. Performance monitoring can help identify potential model degradation in real-world applications.

[0083] Specifically, the identified temperature value output by S4 is continuously compared with the reference temperature value measured by the sensor in the target area, and the average absolute error is calculated.

[0084] S56, Error Feedback: During performance monitoring, if the model's predictions do not meet expectations, the error feedback mechanism will be activated. Error feedback provides information about the model's performance in real-world applications. This feedback can be used to re-evaluate the model, identify problems, and make necessary adjustments.

[0085] Specifically, when the mean absolute error exceeds a preset threshold, it is determined that the model performance has degraded, and an error feedback signal is generated.

[0086] S57, Parameter Update: Based on the error feedback signal, a new round of data acquisition and model training process is initiated. The temperature mapping relationship model in S3 is retrained using the new dataset to update the regression coefficients.

[0087] This invention establishes a multiple linear regression model that considers the influence of environmental factors. By collecting parameters such as thermal imaging pixel values, ambient temperature, and humidity in real time, the model parameters are dynamically optimized, thereby improving the accuracy and robustness of temperature prediction.

[0088] Meanwhile, this invention uses stratified sampling to divide the training set and the validation set, ensuring that the validation set can represent the overall data distribution and avoiding the sample imbalance problem that may be caused by simple random partitioning.

[0089] In addition, the present invention utilizes a mechanism of periodically re-collecting data and updating the model, enabling the temperature recognition model to adapt to dynamic changes in environmental conditions and maintain long-term stable high prediction accuracy.

[0090] The method of the present invention can achieve temperature measurement without contact with the target object. It is simple and quick to operate, and is suitable for temperature monitoring scenarios with a wide range and multiple targets, and has broad application prospects.

[0091] The present invention also discloses a computer-readable storage medium and a computer system. The medium stores a computer program, which, upon execution, performs the method described in any of the preceding claims. A computer system includes a processor and a storage medium, the storage medium storing a computer program, and the processor reading from and running the computer program from the storage medium to perform the method described in any of the preceding claims.

[0092] Those skilled in the art will further appreciate that the various illustrative logic blocks, modules, circuits, and algorithm steps described in conjunction with the embodiments disclosed herein can be implemented as electronic hardware, computer software, or a combination of both. To clearly illustrate this interchangeability between hardware and software, the various illustrative components, blocks, modules, circuits, and steps are described above in a generalized manner in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system. Those skilled in the art may implement the described functionality in different ways for each specific application, but such implementation decisions should not be construed as departing from the scope of the invention.

[0093] The prior description of this disclosure is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to this disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not intended to be limited to the examples and designs described herein, but should be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0094] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for temperature recognition and calculation based on thermal images captured by unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: S1, Data Acquisition: Acquire thermal imaging data and ambient temperature data of the target area, and synchronize the acquired data in time; S2, Process the acquired thermal imaging image data; S3, Establish a mapping model between thermal imaging image data and temperature data; S4 performs temperature prediction and correction; S5. To maintain the accuracy of the model during long-term operation, a dynamic update mechanism is established.

2. The method according to claim 1, characterized in that: In step S1, a high-resolution thermal imaging camera mounted on a drone is used to collect thermal imaging data of the target area, and multiple temperature sensors deployed within the target area are used to collect ambient temperature data.

3. The method according to claim 1, characterized in that, The image data processing in step S2 includes: S21. Image preprocessing: Filter the raw thermal image captured by the UAV to remove noise and perform histogram equalization to enhance image contrast. S22. Target Region Segmentation: The Otsu adaptive threshold segmentation algorithm is used to binarize the preprocessed image and distinguish the foreground target of interest from the irrelevant background. S23. Feature Extraction: For each segmented connected region of the foreground target, calculate its pixel grayscale statistical features, such as the average grayscale value or median value of the region, as pixel feature values ​​representing the temperature level of the region.

4. The method according to claim 1, characterized in that, Step S3 includes the following steps: S31, pairing the pixel feature values ​​of thermal imaging with the actual temperature; S32 uses multiple linear regression to construct a mathematical model for the mapping: Where T is the actual temperature value, These are pixel feature values. ϵ is the regression coefficient, and ϵ is the error term; S33, perform model training and validation, and use cross-validation to evaluate the model's accuracy.

5. The method according to claim 1, characterized in that: In step S4, the established model is used to predict the temperature of the newly acquired thermal image, and the mapping relationship model established in step S3 is updated periodically according to real-time environmental changes to maintain the accuracy of the prediction.

6. The method according to claim 5, characterized in that, The dynamic update mechanism includes the following steps: S51, Data Input: Input the raw data into the model for processing; S52, Model Evaluation: Used to judge the model's performance on the current data; evaluation metrics include accuracy, precision, recall, and F1 score. The evaluation process is carried out using cross-validation to ensure the model's generalization ability. S53, Parameter Update: In order to improve the performance of the model and reduce errors, the parameters of the model are adjusted according to the results of the model evaluation. S54, Model Deployment: Deploy the model, after evaluation and parameter updates, to the production environment for prediction of new data and output the prediction results; S55, Performance Monitoring: Continuously compare the identified temperature value output by S4 with the reference temperature value measured by the sensor in the target area, and calculate the average absolute error. S56, Error Feedback: When the mean absolute error exceeds a preset threshold, it is determined that the model performance has degraded, and an error feedback signal is generated; S57, Parameter Update: Based on the error feedback signal, a new round of data acquisition and model training process is initiated. The temperature mapping relationship model in S3 is retrained using the new dataset to update the regression coefficients.

7. A computer-readable storage medium, characterized in that: The medium contains a computer program, which, when run, performs the method as described in any one of claims 1 to 6.

8. A computer system, characterized in that: It includes a processor and a storage medium, on which a computer program is stored, and the processor reads from the storage medium and runs the computer program to perform the method as described in any one of claims 1 to 6.

Images