Method and device for identifying unmanned aerial vehicle based on infrared temperature measurement

Abstract

The embodiment of the application provides a method and device for identifying a drone based on infrared temperature measurement, the method comprising: obtaining historical drone infrared image data sets, inputting the historical drone infrared image data sets into a convolutional neural network layer and a time sequence processing layer of a preset initial model to extract comprehensive features of the drone, performing parameter tuning operation on the preset initial model according to the comprehensive features of the drone, preset verification set features and a back propagation algorithm, and determining a corresponding drone identification model; receiving an infrared signal, inputting the infrared signal into the drone identification model to perform drone identification operation, determining a corresponding drone identification result, determining a correction result of the corresponding drone identification result according to the drone identification result and a preset environment compensation algorithm, and outputting the correction result of the drone identification result to a display interface for display, so that the efficiency and accuracy of drone identification can be improved.

Description

Technical Field

[0001] This application relates to the field of data processing, specifically to a method and apparatus for identifying unmanned aerial vehicles (UAVs) based on infrared temperature measurement. Background Technology

[0002] With the rapid development of drone technology, drones are increasingly widely used in civilian and commercial fields. However, the widespread use of drones has also brought a series of security and privacy issues, such as illegal intrusion and privacy violations. Therefore, effective detection and monitoring of drones is particularly important.

[0003] Currently, drone detection technologies mainly include radar detection, radio spectrum detection, photoelectric detection, and acoustic detection, but these methods all have certain limitations. Radar has blind spots when detecting close-range targets, making it difficult to effectively detect drones flying at close range; radio spectrum detection cannot detect drones with their radio communication functions turned off; photoelectric detection, especially visible light detection, is greatly affected by weather conditions and has poor detection performance in low visibility conditions; and acoustic detection is insufficient to meet the detection needs of drones targeting distant targets.

[0004] Because current drone detection technologies have various shortcomings and cannot meet the requirements of efficient and accurate drone identification in the field of drone inspection, there is an urgent need for a method that can accurately identify drones in complex environments and is unaffected by the environment. Summary of the Invention

[0005] To address the problems in the prior art, this application provides a method and apparatus for identifying drones based on infrared temperature measurement, which can improve the efficiency and accuracy of drone identification.

[0006] To solve at least one of the above problems, this application provides the following technical solution:

[0007] Firstly, this application provides a method for identifying drones based on infrared temperature measurement, including:

[0008] Receive infrared radiation energy, perform signal processing on the infrared radiation energy, and determine the corresponding infrared signal;

[0009] A historical drone infrared image dataset is acquired. This dataset is then input into a convolutional neural network layer of a preset initial model to extract drone spatial features, determining the corresponding drone static feature vector. The drone static feature vector is then input into a time-series processing layer of the preset initial model to extract drone dynamic features, determining the corresponding drone comprehensive features. Based on the drone comprehensive features, preset validation set features, and a backpropagation algorithm, the preset initial model is optimized to determine the corresponding drone recognition model. The comprehensive features include static and dynamic features. Static features include at least one of drone temperature distribution, drone wing shape, and drone size. Dynamic features include at least one of drone flight speed, drone flight acceleration, and drone flight direction.

[0010] The infrared signal is input into the UAV recognition model to perform UAV recognition operation, and the corresponding UAV recognition result is determined. Based on the UAV recognition result and the preset environmental compensation algorithm, the corresponding UAV recognition result correction result is determined, and the UAV recognition result correction result is output to the display interface for display.

[0011] Further, the receiving of infrared radiation energy and the signal processing operation performed on the infrared radiation energy to determine the corresponding infrared signal include:

[0012] Receive infrared radiation energy, amplify, filter, and sample the infrared radiation energy to determine the corresponding infrared radiation analog signal;

[0013] The infrared radiation analog signal is digitized to determine the corresponding infrared signal.

[0014] Furthermore, the historical UAV infrared image dataset is input into the convolutional neural network layer of a preset initial model to perform UAV spatial feature extraction, determining the corresponding UAV static feature vector, including:

[0015] The historical UAV infrared image dataset is subjected to data preprocessing operations to determine the corresponding UAV infrared image preprocessing dataset. The data preprocessing operations include at least one of data cleaning, noise reduction, image enhancement, and edge detection. The historical UAV infrared image dataset includes at least one of infrared images, radar reflection signals, and flight trajectories.

[0016] The UAV infrared image preprocessing dataset is input into the convolutional neural network layer of a preset initial model to perform UAV spatial feature extraction operations and determine the corresponding UAV static feature vector.

[0017] Further, the step of inputting the UAV infrared image preprocessing dataset into the convolutional neural network layer of a preset initial model to perform UAV spatial feature extraction and determine the corresponding UAV static feature vector includes:

[0018] The UAV infrared image preprocessing dataset is input into the convolutional layer of the convolutional neural network of the preset initial model to perform low-level feature extraction operations and determine the corresponding low-level spatial features, wherein the low-level spatial features include edge features and area features.

[0019] The low-level spatial features are input into the activation layer of the convolutional neural network of the preset initial model to perform complex feature extraction operations and determine the corresponding UAV static feature vector. The static feature vector includes at least one of the following: UAV temperature distribution feature vector, UAV wing shape feature vector, and UAV size feature vector.

[0020] Further, the step of inputting the static feature vector of the UAV into the time series processing layer of the preset initial model to perform UAV dynamic feature extraction operation and determine the corresponding comprehensive UAV features includes:

[0021] The static feature vector of the UAV is input into the input layer of the long short-term recurrent network of the preset initial model. Dynamic feature extraction is performed according to the preset time series analysis algorithm to determine the corresponding dynamic feature vector of the UAV. The dynamic feature vector of the UAV includes at least one of the UAV flight speed feature vector, UAV flight acceleration feature vector, and UAV flight direction feature vector.

[0022] The dynamic feature vector and the static feature vector of the UAV are fused to determine the corresponding comprehensive features of the UAV.

[0023] Further, the step of updating the parameters of the preset initial model based on the comprehensive features of the UAV, the features of the preset verification set, and the backpropagation algorithm to determine the corresponding UAV recognition model includes:

[0024] The corresponding loss function is determined based on the comprehensive characteristics of the UAV and the features of the preset validation set.

[0025] The parameters of the preset initial model are updated based on the minimum loss function and the backpropagation algorithm to determine the corresponding UAV recognition model.

[0026] Further, the step of determining the correction result of the corresponding drone identification result based on the drone identification result and the preset environmental compensation algorithm includes:

[0027] Receive the real ambient temperature and determine the corresponding infrared temperature measurement deviation result based on the real ambient temperature and a preset environmental compensation algorithm;

[0028] The UAV identification result is corrected based on the infrared temperature measurement deviation result to determine the corrected UAV identification result.

[0029] Secondly, this application provides a device for identifying drones based on infrared temperature measurement, comprising:

[0030] An infrared signal processing module is used to receive infrared radiation energy, perform signal processing operations on the infrared radiation energy, and determine the corresponding infrared signal.

[0031] The drone recognition model construction module is used to acquire historical drone infrared image datasets, input the historical drone infrared image datasets into the convolutional neural network layer of a preset initial model to perform drone spatial feature extraction, determine the corresponding drone static feature vectors, input the drone static feature vectors into the time series processing layer of the preset initial model to perform drone dynamic feature extraction, determine the corresponding drone comprehensive features, and perform parameter tuning on the preset initial model based on the drone comprehensive features, preset validation set features, and backpropagation algorithm to determine the corresponding drone recognition model. The comprehensive features include static features and dynamic features. The static features include at least one of drone temperature distribution, drone wing shape, and drone size. The dynamic features include at least one of drone flight speed, drone flight acceleration, and drone flight direction.

[0032] The drone identification and result correction module is used to input the infrared signal into the drone identification model to perform drone identification operation, determine the corresponding drone identification result, determine the corresponding drone identification result correction result based on the drone identification result and a preset environmental compensation algorithm, and output the corrected drone identification result to the display interface for display.

[0033] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method for identifying drones based on infrared temperature measurement.

[0034] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method for identifying unmanned aerial vehicles based on infrared temperature measurement.

[0035] Fifthly, this application provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the method for identifying unmanned aerial vehicles based on infrared temperature measurement.

[0036] As can be seen from the above technical solution, this application provides a method and apparatus for identifying drones based on infrared temperature measurement. By acquiring a historical drone infrared image dataset, the dataset is input into the convolutional neural network layer and time series processing layer of a preset initial model to extract comprehensive drone features. Based on the comprehensive drone features, preset validation set features, and backpropagation algorithm, the preset initial model is parameter-tuned to determine the corresponding drone identification model. Infrared signals are received and input into the drone identification model for drone identification, determining the corresponding drone identification result. Based on the drone identification result and a preset environmental compensation algorithm, a correction result for the drone identification result is determined, and the correction result is output to a display interface for display. This improves the efficiency and accuracy of drone identification. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of this application 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 this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is one of the flowcharts illustrating the method for identifying drones based on infrared temperature measurement in this application embodiment;

[0039] Figure 2 This is a second schematic flowchart of the method for identifying drones based on infrared temperature measurement in the embodiments of this application;

[0040] Figure 3 This is the third flowchart illustrating the method for identifying drones based on infrared temperature measurement in this application.

[0041] Figure 4 This is the fourth flowchart illustrating the method for identifying drones based on infrared temperature measurement in this application.

[0042] Figure 5 This is the fifth flowchart illustrating the method for identifying drones based on infrared temperature measurement in this application.

[0043] Figure 6 This is the sixth flowchart illustrating the method for identifying drones based on infrared temperature measurement in this application.

[0044] Figure 7 This is the seventh flowchart illustrating the method for identifying drones based on infrared temperature measurement in this application.

[0045] Figure 8This is a structural diagram of the device for identifying unmanned aerial vehicles based on infrared temperature measurement in the embodiments of this application;

[0046] Figure 9 This is a schematic diagram of the structure of the electronic device in the embodiments of this application.

[0047] Figure label:

[0048] Electronic device 9600, central processing unit 9100, memory 9140, communication module 9110, input unit 9120, audio processor 9130, display 9160, power supply 9170, buffer memory 9141, application / function storage unit 9142, data storage unit 9143, driver storage unit 9144, antenna 9111, speaker 9131, microphone 9132. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0050] The acquisition, storage, use, and processing of data in this application all comply with the relevant provisions of national laws and regulations.

[0051] Considering the various shortcomings of current drone detection technologies, which fail to meet the requirements of efficient and accurate drone identification in the drone detection field, this application provides a method and apparatus for drone identification based on infrared thermometry. The method involves acquiring a historical drone infrared image dataset, inputting this dataset into the convolutional neural network layer and time series processing layer of a preset initial model for comprehensive drone feature extraction, and then optimizing the parameters of the preset initial model based on the comprehensive drone features, preset validation set features, and a backpropagation algorithm to determine the corresponding drone identification model. Infrared signals are received and input into the drone identification model for drone identification, determining the corresponding drone identification result. Based on the drone identification result and a preset environmental compensation algorithm, a correction result for the drone identification result is determined, and the correction result is output to a display interface for presentation. This improves the efficiency and accuracy of drone identification.

[0052] To improve the efficiency and accuracy of drone identification, this application provides an embodiment of a method for identifying drones based on infrared temperature measurement, see [link to embodiment]. Figure 1The method for identifying drones based on infrared temperature measurement specifically includes the following:

[0053] Step S101: Receive infrared radiation energy, perform signal processing on the infrared radiation energy, and determine the corresponding infrared signal;

[0054] Optionally, in this embodiment, in order to realize the method of identifying UAVs based on infrared temperature measurement, the heat generated by the UAV during flight due to mechanical movement and the operation of electronic components is dissipated into the surrounding environment in the form of infrared radiation. Infrared sensors are used to sense and capture this infrared radiation, and then the infrared radiation energy is converted into infrared signals that can be recognized by machines through signal processing, laying the foundation for subsequent UAV identification algorithms to perform intelligent UAV identification.

[0055] Optionally, in this embodiment, the main component receiving infrared radiation energy is an infrared sensor. A high-sensitivity infrared sensor with anti-interference capabilities is used to receive infrared radiation energy, enabling accurate detection of the drone's presence in complex environments. The sensor's anti-interference design reduces the impact of environmental factors such as temperature changes, direct sunlight, rain, and snow on sensor performance.

[0056] After receiving infrared radiation energy, the infrared radiation energy is first processed as a signal, which includes signal amplification, filtering, and sampling.

[0057] Optionally, signal amplification can be performed to enhance the original infrared image signal. Since infrared radiation signals are often weak, amplification increases the signal strength, making it easier to identify the UAV's infrared characteristics in subsequent processing. The amplification process may adjust the signal amplitude according to preset gain parameters to ensure the signal is not ignored in subsequent processing due to its weakness.

[0058] Optionally, filtering can be performed to remove image noise. Noise can originate from various factors, such as thermal noise from the sensor itself, fluctuations in ambient temperature, and electromagnetic interference. During the filtering process, various filtering algorithms (such as low-pass filtering, high-pass filtering, and band-pass filtering) are applied to suppress noise components while preserving useful information in the image. For example, a low-pass filter can smooth the image and remove high-frequency noise; a high-pass filter can enhance the edge information of the image but may remove some low-frequency background noise.

[0059] Optionally, sampling and collection converts the continuous signal into a discrete signal. In infrared image processing, since the image itself is a discrete signal composed of pixels, this step focuses on extracting single frames from continuously captured infrared video, or downsampling the image to reduce the amount of data. During downsampling, some pixels are selectively discarded, and interpolation algorithms may be applied to reconstruct the remaining pixels to maintain the image's appearance quality.

[0060] Then, the sampled signal is digitized, converting the analog signal into a digital signal. Digitized image data is easier to store, process, and transmit.

[0061] Understandably, infrared images undergo signal amplification, filtering, sampling, and digitization, resulting in significantly improved image quality and providing clearer, more accurate image data for subsequent drone identification and target tracking.

[0062] Step S102: Obtain a historical UAV infrared image dataset. Input the historical UAV infrared image dataset into the convolutional neural network layer of a preset initial model to perform UAV spatial feature extraction, determine the corresponding UAV static feature vector, input the UAV static feature vector into the time series processing layer of the preset initial model to perform UAV dynamic feature extraction, determine the corresponding UAV comprehensive features, and perform parameter tuning on the preset initial model based on the UAV comprehensive features, preset validation set features, and backpropagation algorithm to determine the corresponding UAV recognition model. The comprehensive features include static features and dynamic features. The static features include at least one of UAV temperature distribution, UAV wing shape, and UAV size. The dynamic features include at least one of UAV flight speed, UAV flight acceleration, and UAV flight direction.

[0063] Optionally, in this embodiment, a large proportion and highly diverse historical infrared image dataset is collected to train the initial artificial intelligence model. After training to obtain a small loss function, the parameters of the initial model are updated through the loss function to obtain the UAV recognition model. This model can automatically output the UAV recognition result after receiving the infrared signal passed in step S101.

[0064] Optionally, in order to train the above initial model to obtain the UAV recognition model using a high-quality dataset, a large amount of diverse data collection (infrared images, radar reflection signals, flight trajectories) is required during the data collection phase.

[0065] Optionally, to collect comprehensive and diverse data, the company first collaborated with multiple civilian and military drone manufacturers to acquire flight data for various models and specifications of drones, including infrared images, radar reflection signals, and flight trajectories. Additionally, high-definition infrared cameras were deployed in key areas nationwide to capture drone activity in the sky in real time. Furthermore, to increase the complexity and challenge of the dataset, image data of non-drone targets such as birds, kites, and helicopters were introduced to simulate real-world interference factors.

[0066] Optionally, to obtain a high-quality dataset, preprocessing of the diverse collected data is necessary. The raw data often contains significant amounts of noise, redundant information, and missing values. Therefore, advanced data cleaning and preprocessing techniques are required. First, image processing software is used to denoise the infrared images, removing noise caused by weather, equipment malfunctions, etc. Then, edge detection and image segmentation techniques are used to separate the drone target from the background, followed by normalization to ensure the data input to the model has a consistent format and size. Furthermore, interpolation algorithms are used to impute missing values ​​to ensure data integrity.

[0067] Optionally, after data preprocessing, manual annotation of the dataset is required to train the model, ensuring that the model can accurately learn the correct drone features. Therefore, the data annotation method is crucial to the performance of the model after learning. In this embodiment, to ensure that the model can accurately learn the drone features, the implementation team performed detailed annotation on each image. This included not only basic information such as the drone's location and size, but also further subdivided advanced features such as the drone's type and flight status. To improve the accuracy and efficiency of annotation, semi-automated annotation tools can be used in conjunction with manual review to ensure high-quality annotation results.

[0068] Optionally, the labeled dataset is divided into training, validation, and test sets, with a ratio of 70% training set, 15% validation set, and 15% test set, to ensure that the model can obtain sufficient training data during training and accurately evaluate its performance during validation and testing phases.

[0069] Understandably, the steps described above involve collecting and processing the dataset. During model training, the quality of the dataset determines the performance of the trained model. Next, the model is trained using the processed dataset.

[0070] Optionally, it is necessary to first determine which features the model needs to identify the drone, and then perform feature extraction based on these features. In this embodiment, static and dynamic features for identifying the drone are determined during the feature extraction stage.

[0071] Optional static features include, but are not limited to, drone temperature distribution, drone wing shape, and drone size.

[0072] Optional dynamic features include, but are not limited to, drone flight speed, drone flight acceleration, and drone flight direction.

[0073] Optionally, static feature extraction is performed using a multi-layered CNN model. This involves extracting low-level features (such as edges and areas) and high-level features (such as shape and structure) from labeled infrared images layer by layer, and combining these features hierarchically to form a description of the drone's morphology and size. Simultaneously, considering the characteristics of infrared images, the implementation team emphasized the importance of heat distribution features. Using infrared image processing technology, they extracted the heat peak areas of the drone target and calculated parameters such as temperature gradients and distribution ranges in these areas. These infrared features are crucial for distinguishing the drone from other heat sources (such as buildings, air conditioner units, and ground vehicles).

[0074] By combining the extracted infrared temperature distribution features of the UAV with the above description of the UAV's size and shape, a static and comprehensive description of the UAV target is formed.

[0075] Building upon the above, dynamic features are extracted using an LSTM model combined with time series analysis techniques to capture the dynamic characteristics of the UAV, such as flight trajectory, speed, acceleration, and flight direction. Utilizing timestamp information from radar data and video streams, the team reconstructs the UAV target's flight trajectory and calculates dynamic parameters such as speed, acceleration, and flight direction.

[0076] Then, these spatiotemporal dynamic features are fused with infrared features to form a comprehensive and all-round description of the UAV target.

[0077] Optionally, feature selection and optimization can be performed on the data samples during feature extraction. By using methods such as feature importance assessment and feature dimensionality reduction, redundant and irrelevant features are removed, retaining the most representative feature set. This not only reduces computational cost and improves model training speed but also enhances the model's generalization ability.

[0078] Optionally, after selecting the feature extraction method, an initial model is built using a CNN model and an LSTM model, respectively. The initial model is then trained using a preprocessed dataset. The CNN is responsible for extracting spatial features from the infrared images, while the LSTM is responsible for capturing dynamic features in the time series. This hybrid model can fully utilize information from both infrared images and radar data, improving the accuracy and robustness of UAV identification.

[0079] In this embodiment, during the model training phase, the UAV recognition model can be determined based on the minimum loss function after the initial model training.

[0080] First, before model training, key hyperparameters such as learning rate, batch size, and number of iterations are tuned. Optimal parameter combinations are found using methods such as grid search and random search. Simultaneously, techniques like early stopping can be employed to prevent overfitting. These optimization measures enable the model to converge to the optimal solution more efficiently and stably during training.

[0081] Alternatively, the initial model can employ transfer learning, first pre-training a CNN model on a large and relevant dataset (such as ImageNet), and then transferring the weights of that model to the drone recognition task. By fine-tuning the parameters of the pre-trained model, the implementation team can obtain a high-performance drone recognition model in a relatively short time.

[0082] Secondly, during the formal training phase of the model, the training set of the preprocessed dataset is input into the CNN model and the LATM model in sequence to extract their static and dynamic features.

[0083] Specifically, the training set is first input into the CNN model, and static features are extracted based on the CNN part. The CNN model has multiple feature learning layers, which are used to improve the model's feature representation ability layer by layer from low-level features to high-level features.

[0084] It is understandable that low-level features in an image (such as edges, area, and texture) become more complex and abstract as the number of layers increases, eventually transforming into high-level features (such as shape and structure).

[0085] Specifically, CNN models typically contain the following key layers:

[0086] Convolutional Layer: The convolutional layer is the core of a CNN. It extracts local features from the image by sliding a series of convolutional kernels (also called filters or feature detectors) across the input image, performing local connections and weight sharing operations. In the initial convolutional layers, the model mainly extracts low-level visual features such as edges, corners, and textures. These low-level features are crucial for subsequent image analysis and understanding.

[0087] Activation Layer: Activation layers typically follow convolutional layers and are used to introduce non-linearity, enabling the model to learn more complex patterns. Commonly used activation functions include ReLU (Rectified Linear Unit), which helps the model capture non-linear features in images.

[0088] Pooling Layer: Pooling layers are primarily used for downsampling. By reducing the number of parameters and computational cost, they mitigate the risk of overfitting and improve the model's generalization ability. During pooling, the model subsamples the input feature map, typically selecting the maximum value (max pooling) or average value (average pooling) from the region as the output. This step helps extract the main features of the image while ignoring some less important details.

[0089] Fully Connected Layer: In a fully connected layer, each neuron is connected to all neurons in the previous layer. These layers are typically located at the end of a CNN and are used to non-linearly combine the features extracted by the preceding layers to make the final classification or regression decision.

[0090] By extracting features layer by layer, CNN models can gradually transition from low-level features such as edges and textures to high-level features such as shape and structure, ultimately forming a comprehensive description of the drone target. This hierarchical feature extraction method enables CNNs to demonstrate powerful capabilities in processing image data, particularly achieving remarkable results in tasks such as drone recognition, image classification, and object detection.

[0091] Preferably, pre-trained CNN models such as VGGNet or ResNet can be used as feature extractors. These models have been fully trained on large datasets such as ImageNet and have good feature extraction capabilities. The classification layer of the pre-trained CNN model is removed, and all layers preceding it are retained as the feature extraction part.

[0092] Then, the static feature vector output by the CNN layer is used as input, and the time series data is processed by the LSTM unit to capture the dynamic features during the drone's flight process, and finally output the dynamic feature vector.

[0093] It is worth noting that when the time series processing layer (LSTM layer) receives the static feature output from the convolutional neural network (CNN) and processes the dynamic features of the drone, the original static features exist in an encoded or abstract form. These static features are transformed into a set of feature vectors after passing through the CNN, and these feature vectors contain key spatial information about the drone in the image (such as shape, size, heat distribution, etc.).

[0094] When these feature vectors are input into the LSTM network, the LSTM does not directly "discard" these spatial features, but processes them as part of the time series data. The LSTM network updates and maintains the state of these features in the time series through its internal mechanisms (forget gate, input gate, output gate), thereby enabling it to capture the dynamic changes of the drone in time (such as speed, acceleration, trajectory pattern, orientation, etc.).

[0095] Therefore, during the LSTM process of dynamic features, the original static features are preserved in some form and combined with the dynamic features to form a comprehensive description of the UAV's flight state. This combination enables the model to more accurately identify the UAV and predict its future flight trajectory or behavior.

[0096] Finally, we fuse the obtained static and dynamic feature vectors and input them into a fully connected layer for feature vector mapping to obtain comprehensive features.

[0097] Optionally, after obtaining the comprehensive features of the model output, validation is performed using a validation set to calculate the difference between the model predictions and the actual labels. The loss function measures the model's performance given the input.

[0098] Preferably, during training, the implementation team can employ distributed training techniques to accelerate the training process. They distribute the dataset across multiple GPUs and leverage parallel computing capabilities to update the model's parameters simultaneously. Simultaneously, they can use data augmentation techniques to increase the diversity of the training data and improve the model's generalization ability. After multiple iterations of training, the model gradually converges to the optimal solution and demonstrates good performance on the test set.

[0099] Understandably, after multiple iterations of training, we obtained the minimum loss function. Using the backpropagation algorithm, starting from the output layer, we propagated the gradient of the loss function back to each layer of the network. We then updated the model's weights and biases based on the gradients calculated by the optimization algorithm. Ultimately, we obtained a drone recognition model that can effectively receive infrared signals and identify drones.

[0100] Step S103: Input the infrared signal into the UAV recognition model to perform UAV recognition operation, determine the corresponding UAV recognition result, determine the correction result of the corresponding UAV recognition result based on the UAV recognition result and the preset environmental compensation algorithm, and output the correction result of the UAV recognition result to the display interface for display.

[0101] Optionally, in this implementation, after determining the UAV identification result, correcting the result using environmental compensation technology is an important step to improve the accuracy and stability of the UAV detection system. Environmental compensation technology is mainly used to correct the impact of environmental factors such as changes in ambient temperature, direct sunlight, rain, and snow on the infrared temperature measurement and identification system.

[0102] Optionally, in this embodiment, the environmental compensation technology analyzes the impact of ambient temperature changes on infrared temperature readings based on ambient temperature data and the characteristic curves of the infrared thermometer. Based on the analyzed influence of ambient temperature on infrared temperature measurement, a corresponding environmental compensation algorithm is designed. This process requires additional ambient temperature detection devices to acquire ambient temperature data.

[0103] Optionally, in this embodiment, the collected ambient temperature data is input into an environmental compensation algorithm to calculate the deviation in infrared thermometry readings caused by changes in ambient temperature. Based on the calculated deviation value, the temperature data in the UAV detection results is corrected. The corrected temperature data will more accurately reflect the actual temperature characteristics of the UAV, helping to further improve the accuracy of UAV type and attribute determination.

[0104] For example, the normal operating temperature range of a drone is 40-50℃. If an infrared signal indicates that the temperature of a target object is 45℃, which is within the drone's temperature range, the drone recognition model will identify the target object as a drone and obtain a recognition result. At this time, ambient temperature data is acquired, and the deviation of the infrared temperature measurement reading caused by changes in ambient temperature is used to correct the target object.

[0105] Suppose that, based on ambient temperature data and the characteristic curve of the infrared thermometer, the infrared thermometer reading is 0.5℃ higher for every 1℃ increase in ambient temperature. The baseline ambient temperature is 16℃, and the current ambient temperature is 35℃. Then, applying the environmental compensation algorithm, the measured drone temperature is corrected to 35.5℃ (45℃ - 0.5℃ × (35℃ - 16℃)). At this point, since the target object does not conform to the normal operating temperature range of 40-50℃ for drones, the model should output a result indicating that it is not a drone.

[0106] If, upon obtaining this result, it is found that the target object is indeed not a drone, it indicates that the use of environmental compensation algorithms can help the drone identification model to achieve more accurate identification results.

[0107] If the target is found to be a drone, this type of data can be manually labeled and the drone identification model can be trained and iterated to make the identification results of the drone identification model more accurate.

[0108] In summary, whether the environmental compensation algorithm is used in the correction phase or the retraining phase, it can effectively improve the performance of the UAV recognition model.

[0109] This example demonstrates how this embodiment receives infrared radiation energy and obtains recognition results based on the UAV recognition model, and finally confirms and corrects the recognition results through an environmental compensation algorithm.

[0110] As described above, the method for identifying drones based on infrared temperature measurement provided in this application can obtain a historical drone infrared image dataset, input the historical drone infrared image dataset into the convolutional neural network layer and time series processing layer of a preset initial model to extract comprehensive drone features, perform parameter optimization on the preset initial model based on the comprehensive drone features, preset validation set features, and backpropagation algorithm to determine the corresponding drone identification model; receive infrared signals, input the infrared signals into the drone identification model to perform drone identification operations, determine the corresponding drone identification result, determine the corrected result of the drone identification result based on the drone identification result and a preset environmental compensation algorithm, and output the corrected result of the drone identification result to the display interface for display, thereby improving the efficiency and accuracy of drone identification.

[0111] In one embodiment of the method for identifying drones based on infrared temperature measurement in this application, see [link to relevant documentation]. Figure 2 It can also specifically include the following:

[0112] Step S201: Receive infrared radiation energy, amplify, filter, and sample the infrared radiation energy to determine the corresponding infrared radiation analog signal;

[0113] Step S202: Digitize the infrared radiation analog signal to determine the corresponding infrared signal.

[0114] Optionally, in this embodiment, the main component receiving infrared radiation energy is an infrared sensor. A high-sensitivity infrared sensor with anti-interference capabilities is used to receive infrared radiation energy, enabling accurate detection of the drone's presence in complex environments. The sensor's anti-interference design reduces the impact of environmental factors such as temperature changes, direct sunlight, rain, and snow on sensor performance.

[0115] After receiving infrared radiation energy, the infrared radiation energy is first processed as a signal, which includes signal amplification, filtering, and sampling.

[0116] Optionally, signal amplification can be performed to enhance the original infrared image signal. Since infrared radiation signals are often weak, amplification increases the signal strength, making it easier to identify the UAV's infrared characteristics in subsequent processing. The amplification process may adjust the signal amplitude according to preset gain parameters to ensure the signal is not ignored in subsequent processing due to its weakness.

[0117] Optionally, filtering can be performed to remove image noise. Noise can originate from various factors, such as thermal noise from the sensor itself, fluctuations in ambient temperature, and electromagnetic interference. During the filtering process, various filtering algorithms (such as low-pass filtering, high-pass filtering, and band-pass filtering) are applied to suppress noise components while preserving useful information in the image. For example, a low-pass filter can smooth the image and remove high-frequency noise; a high-pass filter can enhance the edge information of the image but may remove some low-frequency background noise.

[0118] Optionally, sampling and collection converts the continuous signal into a discrete signal. In infrared image processing, since the image itself is a discrete signal composed of pixels, this step focuses on extracting single frames from continuously captured infrared video, or downsampling the image to reduce the amount of data. During downsampling, some pixels are selectively discarded, and interpolation algorithms may be applied to reconstruct the remaining pixels to maintain the image's appearance quality.

[0119] Then, the sampled signal is digitized, converting the analog signal into a digital signal. Digitized image data is easier to store, process, and transmit.

[0120] Through step S202, this embodiment obtains an infrared signal that has undergone signal amplification, filtering, sampling, and digitization, providing clearer and more accurate image data for subsequent UAV identification and target tracking.

[0121] In one embodiment of the method for identifying drones based on infrared temperature measurement in this application, see [link to relevant documentation]. Figure 3 It can also specifically include the following:

[0122] Step S301: Perform data preprocessing operations on the historical UAV infrared image dataset to determine the corresponding UAV infrared image preprocessing dataset. The data preprocessing operations include at least one of data cleaning, noise reduction, image enhancement, and edge detection. The historical UAV infrared image dataset includes at least one of infrared images, radar reflection signals, and flight trajectories.

[0123] Step S302: Input the UAV infrared image preprocessing dataset into the convolutional neural network layer of the preset initial model to perform UAV spatial feature extraction operation and determine the corresponding UAV static feature vector.

[0124] Optionally, to collect comprehensive and diverse historical drone infrared imagery data, we first collaborated with multiple civilian and military drone manufacturers to acquire flight data for various models and specifications of drones, including infrared images, radar reflection signals, and flight trajectories. Additionally, we deployed high-definition infrared cameras in key areas nationwide to capture drone activity in the sky in real time. Furthermore, to increase the complexity and challenge of the dataset, image data of non-drone targets such as birds, kites, and helicopters were introduced to simulate real-world interference factors.

[0125] Optionally, to obtain a high-quality dataset, preprocessing of the diverse collected data is necessary. The raw data often contains significant amounts of noise, redundant information, and missing values. Therefore, advanced data cleaning and preprocessing techniques are required. First, image processing software is used to denoise the infrared images, removing noise caused by weather, equipment malfunctions, etc. Then, edge detection and image segmentation techniques are used to separate the drone target from the background, followed by normalization to ensure the data input to the model has a consistent format and size. Furthermore, interpolation algorithms are used to impute missing values ​​to ensure data integrity.

[0126] Understandably, the steps described above involve collecting and processing the dataset. During model training, the quality of the dataset determines the performance of the trained model. Next, the model is trained using the processed dataset.

[0127] Through step S302, this embodiment realizes the collection and processing of historical infrared datasets of UAVs, increasing their quality while maintaining their diversity, and laying a foundation for subsequent model training to obtain an efficient UAV recognition model.

[0128] In one embodiment of the method for identifying drones based on infrared temperature measurement in this application, see [link to relevant documentation]. Figure 4 It can also specifically include the following:

[0129] Step S401: Input the UAV infrared image preprocessing dataset into the convolutional layer of the convolutional neural network of the preset initial model to perform low-level feature extraction operation and determine the corresponding low-level spatial features, wherein the low-level spatial features include edge features and area features;

[0130] Step S402: Input the low-level spatial features into the activation layer of the convolutional neural network of the preset initial model, perform complex feature extraction operations, and determine the corresponding UAV static feature vector, wherein the static feature vector includes at least one of the UAV temperature distribution feature vector, UAV wing shape feature vector, and UAV size feature vector.

[0131] Optionally, a multi-layered CNN model can be used to extract low-level features (such as edges and areas) and high-level features (such as shape and structure) from labeled infrared images layer by layer, and combine these features in a hierarchical manner to form a feature description of the drone's wing shape and size.

[0132] Meanwhile, considering the characteristics of infrared images, the implementation team emphasized the importance of heat distribution features. Using infrared image processing technology, they extracted the heat peak areas of the UAV target and calculated parameters such as temperature gradients and distribution ranges in these areas, obtaining the UAV's temperature distribution characteristics. These infrared features are crucial for distinguishing UAVs from other heat sources (such as buildings, air conditioner outdoor units, and ground vehicles).

[0133] By combining the extracted infrared temperature distribution features of the UAV with the aforementioned UAV size and wing morphology features, a static and comprehensive description of the UAV target is formed.

[0134] It is understandable that low-level features in an image (such as edges, area, and texture) become more complex and abstract as the number of layers increases, eventually transforming into high-level features (such as shape and structure).

[0135] Specifically, CNN models typically contain the following key layers:

[0136] Convolutional Layer: The convolutional layer is the core of a CNN. It extracts local features from the image by sliding a series of convolutional kernels (also called filters or feature detectors) across the input image, performing local connections and weight sharing operations. In the initial convolutional layers, the model mainly extracts low-level visual features such as edges, corners, and textures. These low-level features are crucial for subsequent image analysis and understanding.

[0137] Activation Layer: Activation layers typically follow convolutional layers and are used to introduce non-linearity, enabling the model to learn more complex patterns. Commonly used activation functions include ReLU (Rectified Linear Unit), which helps the model capture non-linear features in images.

[0138] Pooling Layer: Pooling layers are primarily used for downsampling. By reducing the number of parameters and computational cost, they mitigate the risk of overfitting and improve the model's generalization ability. During pooling, the model subsamples the input feature map, typically selecting the maximum value (max pooling) or average value (average pooling) from the region as the output. This step helps extract the main features of the image while ignoring some less important details.

[0139] Fully Connected Layer: In a fully connected layer, each neuron is connected to all neurons in the previous layer. These layers are typically located at the end of a CNN and are used to non-linearly combine the features extracted by the preceding layers to make the final classification or regression decision.

[0140] By extracting features layer by layer, CNN models can gradually transition from low-level features such as edges and textures to high-level features such as shape and structure, ultimately forming a comprehensive description of the drone target. This hierarchical feature extraction method enables CNNs to demonstrate powerful capabilities in processing image data, particularly achieving remarkable results in tasks such as drone recognition, image classification, and object detection.

[0141] Preferably, pre-trained CNN models such as VGGNet or ResNet can be used as feature extractors. These models have been fully trained on large datasets such as ImageNet and have good feature extraction capabilities. The classification layer of the pre-trained CNN model is removed, and all layers preceding it are retained as the feature extraction part.

[0142] Through step S402, this embodiment successfully extracted the static features for UAV identification, laying the foundation for subsequent training of the UAV identification model.

[0143] In one embodiment of the method for identifying drones based on infrared temperature measurement in this application, see [link to relevant documentation]. Figure 5 It can also specifically include the following:

[0144] Step S501: Input the static feature vector of the UAV into the input layer of the long short-term recurrent network of the preset initial model, and perform dynamic feature extraction operation according to the preset time series analysis algorithm to determine the corresponding UAV dynamic feature vector. The UAV dynamic feature vector includes at least one of the following: UAV flight speed feature vector, UAV flight acceleration feature vector, and UAV flight direction feature vector.

[0145] Step S502: Perform a fusion operation on the UAV dynamic feature vector and the UAV static feature vector to determine the corresponding comprehensive UAV features.

[0146] Optionally, the static feature vector output by the CNN layer can be used as input, and the time series data can be processed by the LSTM unit to capture the dynamic features during the flight of the drone, and finally output the dynamic feature vector.

[0147] It is worth noting that when the time series processing layer (LSTM layer) receives the static feature output from the convolutional neural network (CNN) and processes the dynamic features of the drone, the original static features exist in an encoded or abstract form. These static features are transformed into a set of feature vectors after passing through the CNN, and these feature vectors contain key spatial information about the drone in the image (such as shape, size, heat distribution, etc.).

[0148] When these feature vectors are input into the LSTM network, the LSTM does not directly "discard" these spatial features, but processes them as part of the time series data. The LSTM network updates and maintains the state of these features in the time series through its internal mechanisms (forget gate, input gate, output gate), thereby enabling it to capture the dynamic changes of the drone in time (such as speed, acceleration, trajectory pattern, orientation, etc.).

[0149] Therefore, during the LSTM process of dynamic features, the original static features are preserved in some form and combined with the dynamic features to form a comprehensive description of the UAV's flight state. This combination enables the model to more accurately identify the UAV and predict its future flight trajectory or behavior.

[0150] Finally, we fuse the obtained static and dynamic feature vectors and input them into a fully connected layer for feature vector mapping to obtain comprehensive features.

[0151] Through step S502, this embodiment successfully determined the dynamic feature vector and comprehensive features, laying the foundation for subsequent training of the UAV recognition model.

[0152] In one embodiment of the method for identifying drones based on infrared temperature measurement in this application, see [link to relevant documentation]. Figure 6 It can also specifically include the following:

[0153] Step S601: Determine the corresponding loss function based on the comprehensive features of the UAV and the features of the preset validation set;

[0154] Step S602: Update the parameters of the preset initial model according to the minimum loss function and backpropagation algorithm to determine the corresponding UAV recognition model.

[0155] Optionally, after obtaining the comprehensive features of the model output, validation is performed using a validation set to calculate the difference between the model predictions and the actual labels. The loss function measures the model's performance given the input.

[0156] Preferably, during training, the implementation team can employ distributed training techniques to accelerate the training process. They distribute the dataset across multiple GPUs and leverage parallel computing capabilities to update the model's parameters simultaneously. Simultaneously, they can use data augmentation techniques to increase the diversity of the training data and improve the model's generalization ability. After multiple iterations of training, the model gradually converges to the optimal solution and demonstrates good performance on the test set.

[0157] Understandably, after multiple iterations of training, we obtained the minimum loss function. Using the backpropagation algorithm, starting from the output layer, we propagated the gradient of the loss function back to each layer of the network. We then updated the model's weights and biases based on the gradients calculated by the optimization algorithm. Ultimately, we obtained a drone recognition model that can effectively receive infrared signals and identify drones.

[0158] Through step S602, this embodiment successfully obtained the UAV identification model, laying the foundation for subsequent output of UAV identification results.

[0159] In one embodiment of the method for identifying drones based on infrared temperature measurement in this application, see [link to relevant documentation]. Figure 7 It can also specifically include the following:

[0160] Step S701: Receive the actual ambient temperature, and determine the corresponding infrared temperature measurement deviation result based on the actual ambient temperature and the preset environmental compensation algorithm;

[0161] Step S702: Correct the UAV identification result based on the infrared temperature measurement deviation result, and determine the corrected result of the corresponding UAV identification result.

[0162] Optionally, in this implementation, after determining the UAV identification result, correcting the result using environmental compensation technology is an important step to improve the accuracy and stability of the UAV detection system. Environmental compensation technology is mainly used to correct the impact of environmental factors such as changes in ambient temperature, direct sunlight, rain, and snow on the infrared temperature measurement and identification system.

[0163] Optionally, in this embodiment, the environmental compensation technology analyzes the impact of ambient temperature changes on infrared temperature readings based on ambient temperature data and the characteristic curves of the infrared thermometer. Based on the analyzed influence of ambient temperature on infrared temperature measurement, a corresponding environmental compensation algorithm is designed. This process requires additional ambient temperature detection devices to acquire ambient temperature data.

[0164] Optionally, in this embodiment, the collected ambient temperature data is input into an environmental compensation algorithm to calculate the deviation in infrared thermometry readings caused by changes in ambient temperature. Based on the calculated deviation value, the temperature data in the UAV detection results is corrected. The corrected temperature data will more accurately reflect the actual temperature characteristics of the UAV, helping to further improve the accuracy of UAV type and attribute determination.

[0165] For example, the normal operating temperature range of a drone is 40-50℃. If an infrared signal indicates that the temperature of a target object is 45℃, which is within the drone's temperature range, the drone recognition model will identify the target object as a drone and obtain a recognition result. At this time, ambient temperature data is acquired, and the deviation of the infrared temperature measurement reading caused by changes in ambient temperature is used to correct the target object.

[0166] Suppose that, based on ambient temperature data and the characteristic curve of the infrared thermometer, for every 1°C increase in ambient temperature, the infrared thermometer reading will be 0.5°C higher. The baseline ambient temperature is 16°C, and the current ambient temperature is 35°C. Then, applying the environmental compensation algorithm, the measured drone temperature is corrected to 35.5°C (45°C - 0.5°C × (35°C - 16°C)). At this point, since the target object does not conform to the normal operating temperature range of 40-50°C for drones, the model should output a result indicating that it is not a drone.

[0167] If, upon obtaining this result, it is found that the target object is indeed not a drone, it indicates that the use of environmental compensation algorithms can help the drone identification model to achieve more accurate identification results.

[0168] If the target is found to be a drone, this type of data can be manually labeled and the drone identification model can be trained and iterated to make the identification results of the drone identification model more accurate.

[0169] In summary, whether the environmental compensation algorithm is used in the correction phase or the retraining phase, it can effectively improve the performance of the UAV recognition model.

[0170] Through step S702, this embodiment successfully performs environmental compensation correction on the results of the UAV recognition model, thereby achieving more accurate UAV recognition.

[0171] To improve the efficiency and accuracy of drone identification, this application provides an embodiment of an infrared temperature measurement-based drone identification device for implementing all or part of the aforementioned method for drone identification. See [link to embodiment]. Figure 8 The device for identifying drones based on infrared temperature measurement specifically includes the following components:

[0172] Infrared signal processing module 10 is used to receive infrared radiation energy, perform signal processing operations on the infrared radiation energy, and determine the corresponding infrared signal.

[0173] The UAV recognition model construction module 20 is used to acquire a historical UAV infrared image dataset, input the historical UAV infrared image dataset into the convolutional neural network layer of a preset initial model to perform UAV spatial feature extraction, determine the corresponding UAV static feature vector, input the UAV static feature vector into the time series processing layer of the preset initial model to perform UAV dynamic feature extraction, determine the corresponding UAV comprehensive features, and perform parameter tuning on the preset initial model based on the UAV comprehensive features, preset validation set features, and backpropagation algorithm to determine the corresponding UAV recognition model. The comprehensive features include static features and dynamic features. The static features include at least one of UAV temperature distribution, UAV wing shape, and UAV size. The dynamic features include at least one of UAV flight speed, UAV flight acceleration, and UAV flight direction.

[0174] The drone identification and result correction module 30 is used to input the infrared signal into the drone identification model to perform drone identification operation, determine the corresponding drone identification result, determine the correction result of the corresponding drone identification result based on the drone identification result and the preset environmental compensation algorithm, and output the correction result of the drone identification result to the display interface for display.

[0175] As described above, the device for identifying drones based on infrared temperature measurement provided in this application can acquire historical drone infrared image datasets, input the historical drone infrared image datasets into the convolutional neural network layer and time series processing layer of a preset initial model to extract comprehensive drone features, perform parameter optimization on the preset initial model based on the comprehensive drone features, preset validation set features, and backpropagation algorithm to determine the corresponding drone identification model; receive infrared signals, input the infrared signals into the drone identification model to perform drone identification operations, determine the corresponding drone identification result, determine the correction result of the corresponding drone identification result based on the drone identification result and a preset environmental compensation algorithm, and output the correction result of the drone identification result to the display interface for display, thereby improving the efficiency and accuracy of drone identification.

[0176] From a hardware perspective, in order to improve the efficiency and accuracy of drone identification, this application provides an embodiment of an electronic device for implementing all or part of the method for identifying drones based on infrared temperature measurement. The electronic device specifically includes the following components:

[0177] The system comprises a processor, memory, a communications interface, and a bus; wherein the processor, memory, and communications interface communicate with each other via the bus; the communications interface is used to realize information transmission between the method for identifying drones based on infrared temperature measurement and the core business system, user terminals, and related databases and other related devices; the logic controller can be a desktop computer, tablet computer, or mobile terminal, etc., and this embodiment is not limited to these. In this embodiment, the logic controller can be implemented with reference to the embodiments of the method for identifying drones based on infrared temperature measurement in the previous embodiment, and the contents of the embodiments of the method for identifying drones based on infrared temperature measurement are incorporated herein, and repeated details will not be described again.

[0178] It is understood that the user terminal may include smartphones, tablet computers, network set-top boxes, portable computers, desktop computers, personal digital assistants (PDAs), in-vehicle devices, smart wearable devices, etc. Among these, the smart wearable devices may include smart glasses, smartwatches, smart bracelets, etc.

[0179] In practical applications, the method for identifying drones based on infrared temperature measurement can be partially executed on the electronic device side as described above, or all operations can be completed in the client device. The choice can be made based on the processing power of the client device and the limitations of the user's usage scenario. This application does not impose any limitations on this. If all operations are completed in the client device, the client device may further include a processor.

[0180] The aforementioned client device may have a communication module (i.e., a communication unit) that can communicate with a remote server to achieve data transmission with the server. The server may include a server on the task scheduling center side; in other implementation scenarios, it may also include a server on an intermediate platform, such as a server on a third-party server platform that has a communication link with the task scheduling center server. The server may include a single computer device, a server cluster consisting of multiple servers, or a distributed server structure.

[0181] Figure 9 This is a schematic block diagram illustrating the system configuration of the electronic device 9600 according to an embodiment of this application. Figure 9 As shown, the electronic device 9600 may include a central processing unit 9100 and a memory 9140; the memory 9140 is coupled to the central processing unit 9100. It is worth noting that... Figure 9 This is an example; other types of structures can also be used to supplement or replace this structure to achieve telecommunications functions or other functions.

[0182] In one embodiment, the method for identifying drones based on infrared temperature measurement can be integrated into a central processing unit 9100. The central processing unit 9100 can be configured to perform the following control:

[0183] Step S101: Receive infrared radiation energy, perform signal processing on the infrared radiation energy, and determine the corresponding infrared signal;

[0184] Step S102: Obtain a historical UAV infrared image dataset. Input the historical UAV infrared image dataset into the convolutional neural network layer of a preset initial model to perform UAV spatial feature extraction, determine the corresponding UAV static feature vector, input the UAV static feature vector into the time series processing layer of the preset initial model to perform UAV dynamic feature extraction, determine the corresponding UAV comprehensive features, and perform parameter tuning on the preset initial model based on the UAV comprehensive features, preset validation set features, and backpropagation algorithm to determine the corresponding UAV recognition model. The comprehensive features include static features and dynamic features. The static features include at least one of UAV temperature distribution, UAV wing shape, and UAV size. The dynamic features include at least one of UAV flight speed, UAV flight acceleration, and UAV flight direction.

[0185] Step S103: Input the infrared signal into the UAV recognition model to perform UAV recognition operation, determine the corresponding UAV recognition result, determine the correction result of the corresponding UAV recognition result based on the UAV recognition result and the preset environmental compensation algorithm, and output the correction result of the UAV recognition result to the display interface for display.

[0186] As described above, the electronic device provided in this application embodiment acquires a historical drone infrared image dataset, inputs the historical drone infrared image dataset into the convolutional neural network layer and time series processing layer of a preset initial model to extract comprehensive drone features, performs parameter tuning on the preset initial model based on the comprehensive drone features, preset validation set features, and backpropagation algorithm, and determines the corresponding drone recognition model; receives infrared signals, inputs the infrared signals into the drone recognition model to perform drone recognition operation, determines the corresponding drone recognition result, determines the correction result of the corresponding drone recognition result based on the drone recognition result and a preset environmental compensation algorithm, and outputs the correction result of the drone recognition result to the display interface for display, thereby improving the efficiency and accuracy of drone recognition.

[0187] In another embodiment, the method for identifying drones based on infrared temperature measurement can be configured separately from the central processing unit 9100. For example, the method for identifying drones based on infrared temperature measurement can be configured as a chip connected to the central processing unit 9100, and the function of the method for identifying drones based on infrared temperature measurement can be realized through the control of the central processing unit.

[0188] like Figure 9 As shown, the electronic device 9600 may further include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It is worth noting that the electronic device 9600 does not necessarily need to include these components. Figure 9 All components shown; in addition, the electronic device 9600 may also include Figure 9 For components not shown, please refer to existing technology.

[0189] like Figure 9 As shown, the central processing unit 9100, sometimes also referred to as a controller or operating control, may include a microprocessor or other processor device and / or logic device, which receives inputs and controls the operation of various components of the electronic device 9600.

[0190] The memory 9140 may be, for example, one or more of a cache, flash memory, hard drive, removable media, volatile memory, non-volatile memory, or other suitable devices. It may store the aforementioned failure-related information, and also store a program for executing that information. The central processing unit 9100 may execute the program stored in the memory 9140 to perform information storage or processing, etc.

[0191] Input unit 9120 provides input to central processing unit 9100. Input unit 9120 may be, for example, a keypad or touch input device. Power supply 9170 provides power to electronic device 9600. Display 9160 displays images and text. Display may be, for example, an LCD display, but is not limited thereto.

[0192] The memory 9140 can be a solid-state memory, such as a read-only memory (ROM), random access memory (RAM), a SIM card, etc. It can also be a memory that retains information even when power is off, can be selectively erased, and contains more data; examples of this type of memory are sometimes referred to as EPROMs. The memory 9140 can also be some other type of device. The memory 9140 includes a buffer memory 9141 (sometimes referred to as a buffer). The memory 9140 may include an application / function storage unit 9142 for storing application programs and function programs or processes for executing the operation of the electronic device 9600 via the central processing unit 9100.

[0193] The memory 9140 may also include a data storage unit 9143 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 9144 of the memory 9140 may include various drivers for the electronic device for communication functions and / or for performing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0194] The communication module 9110 is a transmitter / receiver that sends and receives signals via the antenna 9111. The communication module 9110 (transmitter / receiver) is coupled to the central processing unit 9100 to provide input signals and receive output signals, which is the same as in a conventional mobile communication terminal.

[0195] Based on different communication technologies, multiple communication modules 9110 can be configured in the same electronic device, such as cellular network modules, Bluetooth modules, and / or wireless LAN modules. The communication module (transmitter / receiver) 9110 is also coupled to a speaker 9131 and a microphone 9132 via an audio processor 9130 to provide audio output via the speaker 9131 and receive audio input from the microphone 9132, thereby realizing typical telecommunications functions. The audio processor 9130 may include any suitable buffer, decoder, amplifier, etc. Additionally, the audio processor 9130 is also coupled to a central processing unit 9100, enabling on-device recording via the microphone 9132 and on-device playback of stored audio via the speaker 9131.

[0196] Embodiments of this application also provide a computer-readable storage medium capable of implementing all steps of the method for infrared temperature measurement and identification of a drone with a server or client as the execution subject in the above embodiments. The computer-readable storage medium stores a computer program that, when executed by a processor, implements all steps of the method for infrared temperature measurement and identification of a drone with a server or client as the execution subject in the above embodiments. For example, when the processor executes the computer program, it implements the following steps:

[0197] Step S101: Receive infrared radiation energy, perform signal processing on the infrared radiation energy, and determine the corresponding infrared signal;

[0198] Step S102: Obtain a historical UAV infrared image dataset. Input the historical UAV infrared image dataset into the convolutional neural network layer of a preset initial model to perform UAV spatial feature extraction, determine the corresponding UAV static feature vector, input the UAV static feature vector into the time series processing layer of the preset initial model to perform UAV dynamic feature extraction, determine the corresponding UAV comprehensive features, and perform parameter tuning on the preset initial model based on the UAV comprehensive features, preset validation set features, and backpropagation algorithm to determine the corresponding UAV recognition model. The comprehensive features include static features and dynamic features. The static features include at least one of UAV temperature distribution, UAV wing shape, and UAV size. The dynamic features include at least one of UAV flight speed, UAV flight acceleration, and UAV flight direction.

[0199] Step S103: Input the infrared signal into the UAV recognition model to perform UAV recognition operation, determine the corresponding UAV recognition result, determine the correction result of the corresponding UAV recognition result based on the UAV recognition result and the preset environmental compensation algorithm, and output the correction result of the UAV recognition result to the display interface for display.

[0200] As described above, the computer-readable storage medium provided in this application embodiment acquires a historical UAV infrared image dataset, inputs the historical UAV infrared image dataset into the convolutional neural network layer and time series processing layer of a preset initial model to extract comprehensive UAV features, performs parameter tuning on the preset initial model based on the comprehensive UAV features, preset validation set features, and backpropagation algorithm, and determines the corresponding UAV recognition model; receives infrared signals, inputs the infrared signals into the UAV recognition model to perform UAV recognition operation, determines the corresponding UAV recognition result, determines the corresponding UAV recognition result correction result based on the UAV recognition result and a preset environment compensation algorithm, and outputs the corrected UAV recognition result to a display interface for display, thereby improving the efficiency and accuracy of UAV recognition.

[0201] Embodiments of this application also provide a computer program product capable of implementing all steps of the method for identifying an infrared temperature-measuring drone, where the execution subject is a server or client, as described in the above embodiments. When executed by a processor, this computer program / instruction implements the steps of the method for identifying an infrared temperature-measuring drone. For example, the computer program / instruction implements the following steps:

[0202] Step S101: Receive infrared radiation energy, perform signal processing on the infrared radiation energy, and determine the corresponding infrared signal;

[0203] Step S102: Obtain a historical UAV infrared image dataset. Input the historical UAV infrared image dataset into the convolutional neural network layer of a preset initial model to perform UAV spatial feature extraction, determine the corresponding UAV static feature vector, input the UAV static feature vector into the time series processing layer of the preset initial model to perform UAV dynamic feature extraction, determine the corresponding UAV comprehensive features, and perform parameter tuning on the preset initial model based on the UAV comprehensive features, preset validation set features, and backpropagation algorithm to determine the corresponding UAV recognition model. The comprehensive features include static features and dynamic features. The static features include at least one of UAV temperature distribution, UAV wing shape, and UAV size. The dynamic features include at least one of UAV flight speed, UAV flight acceleration, and UAV flight direction.

[0204] Step S103: Input the infrared signal into the UAV recognition model to perform UAV recognition operation, determine the corresponding UAV recognition result, determine the correction result of the corresponding UAV recognition result based on the UAV recognition result and the preset environmental compensation algorithm, and output the correction result of the UAV recognition result to the display interface for display.

[0205] As described above, the computer program product provided in this application embodiment acquires a historical UAV infrared image dataset, inputs the historical UAV infrared image dataset into the convolutional neural network layer and time series processing layer of a preset initial model to extract comprehensive UAV features, performs parameter tuning on the preset initial model based on the comprehensive UAV features, preset validation set features, and backpropagation algorithm, and determines the corresponding UAV recognition model; receives infrared signals, inputs the infrared signals into the UAV recognition model to perform UAV recognition operation, determines the corresponding UAV recognition result, determines the corresponding UAV recognition result correction result based on the UAV recognition result and a preset environmental compensation algorithm, and outputs the corrected UAV recognition result to the display interface for display, thereby improving the efficiency and accuracy of UAV recognition.

[0206] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0207] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (devices), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0208] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0209] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0210] Specific embodiments have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A method for identifying a drone based on infrared temperature measurement, characterized in that, The method comprises: receiving infrared radiation energy, performing signal processing operations on the infrared radiation energy, and determining a corresponding infrared signal; obtaining a historical unmanned aerial vehicle infrared image dataset, performing data preprocessing operations on the historical unmanned aerial vehicle infrared image dataset to determine a corresponding unmanned aerial vehicle infrared image preprocessed dataset, inputting the unmanned aerial vehicle infrared image preprocessed dataset into a convolutional layer of a convolutional neural network of a preset initial model, performing low-level feature extraction operations to determine a corresponding low-level spatial feature, inputting the low-level spatial feature into an activation layer of the convolutional neural network of the preset initial model, performing complex feature extraction operations to determine a corresponding unmanned aerial vehicle static feature vector, inputting the unmanned aerial vehicle static feature vector into a time series processing layer of the preset initial model to perform unmanned aerial vehicle dynamic feature extraction operations to determine a corresponding unmanned aerial vehicle comprehensive feature, updating parameters of the preset initial model according to the unmanned aerial vehicle comprehensive feature, a preset validation set feature, and a back propagation algorithm, and determining a corresponding unmanned aerial vehicle recognition model, wherein the data preprocessing operations include at least one of data cleaning, denoising, image enhancement, and edge detection, the historical unmanned aerial vehicle infrared image dataset includes at least one of an infrared image, a radar reflection signal, and a flight trajectory, the low-level spatial feature includes an edge feature and an area feature, the comprehensive feature includes a static feature and a dynamic feature, the static feature includes at least one of an unmanned aerial vehicle temperature distribution, an unmanned aerial vehicle wing shape, and an unmanned aerial vehicle size, the dynamic feature includes at least one of an unmanned aerial vehicle flight speed, an unmanned aerial vehicle flight acceleration, and an unmanned aerial vehicle flight direction, and the unmanned aerial vehicle temperature distribution is obtained by using infrared image processing technology to extract a heat peak area of an unmanned aerial vehicle target, and performing calculation operations on the heat peak area obtained after the extraction to determine corresponding temperature gradient parameters and distribution range parameters; inputting the infrared signal into the unmanned aerial vehicle recognition model to perform unmanned aerial vehicle recognition operations to determine a corresponding unmanned aerial vehicle recognition result, receiving an actual environment temperature, determining a corresponding infrared temperature measurement deviation result according to the actual environment temperature and a preset environment compensation algorithm, performing correction operations on the unmanned aerial vehicle recognition result according to the infrared temperature measurement deviation result to determine a correction result of the unmanned aerial vehicle recognition result, and outputting the correction result of the unmanned aerial vehicle recognition result to a display interface for display. 2.The method for identifying a UAV based on infrared temperature measurement according to claim 1, characterized in that, The receiving infrared radiation energy, performing signal processing operations on the infrared radiation energy, and determining a corresponding infrared signal comprises: receiving infrared radiation energy, performing signal amplification, filtering, and sampling operations on the infrared radiation energy, and determining a corresponding infrared radiation analog signal; performing digitization operations on the infrared radiation analog signal to determine a corresponding infrared signal. 3.The method of identifying a drone based on infrared temperature measurement according to claim 1, characterized in that, The inputting the unmanned aerial vehicle static feature vector into a time series processing layer of a preset initial model to perform unmanned aerial vehicle dynamic feature extraction operations to determine a corresponding unmanned aerial vehicle comprehensive feature comprises: The unmanned aerial vehicle static feature vector is input into an input layer of a long short-term recurrent network of a preset initial model, dynamic feature extraction is performed according to a preset time sequence analysis algorithm, and a corresponding unmanned aerial vehicle dynamic feature vector is determined, wherein the unmanned aerial vehicle dynamic feature vector includes at least one of an unmanned aerial vehicle flight speed feature vector, an unmanned aerial vehicle flight acceleration feature vector, and an unmanned aerial vehicle flight direction feature vector; The unmanned aerial vehicle dynamic feature vector and the unmanned aerial vehicle static feature vector are fused to determine a corresponding unmanned aerial vehicle comprehensive feature. 4.The method of identifying a drone based on infrared temperature measurement according to claim 1, wherein, The parameters of the preset initial model are updated according to the unmanned aerial vehicle comprehensive feature, a preset verification set feature, and a back propagation algorithm to determine a corresponding unmanned aerial vehicle recognition model, including: A loss function is determined according to the unmanned aerial vehicle comprehensive feature and the preset verification set feature; The parameters of the preset initial model are updated according to the smallest loss function and the back propagation algorithm to determine the corresponding unmanned aerial vehicle recognition model.

5. A device for identifying unmanned aerial vehicles based on infrared temperature measurement, characterized in that, The device includes: An infrared signal processing module for receiving infrared radiation energy, performing signal processing on the infrared radiation energy, and determining a corresponding infrared signal; An unmanned aerial vehicle recognition model construction module for obtaining a historical unmanned aerial vehicle infrared image data set, performing data preprocessing on the historical unmanned aerial vehicle infrared image data set to determine a corresponding unmanned aerial vehicle infrared image preprocessing data set, inputting the unmanned aerial vehicle infrared image preprocessing data set into a convolution layer of a convolutional neural network of a preset initial model to perform low-level feature extraction to determine a corresponding low-level spatial feature, inputting the low-level spatial feature into an activation layer of the convolutional neural network of the preset initial model to perform complex feature extraction to determine a corresponding unmanned aerial vehicle static feature vector, inputting the unmanned aerial vehicle static feature vector into a time sequence processing layer of the preset initial model to perform unmanned aerial vehicle dynamic feature extraction to determine a corresponding unmanned aerial vehicle comprehensive feature, and performing parameter optimization on the preset initial model according to the unmanned aerial vehicle comprehensive feature, a preset verification set feature, and a back propagation algorithm to determine a corresponding unmanned aerial vehicle recognition model, wherein the data preprocessing includes at least one of data cleaning, denoising, image enhancement, and edge detection, the historical unmanned aerial vehicle infrared image data set includes at least one of an infrared image, a radar reflection signal, and a flight trajectory, the low-level spatial feature includes an edge feature and an area feature, the comprehensive feature includes a static feature and a dynamic feature, the static feature includes at least one of an unmanned aerial vehicle temperature distribution, an unmanned aerial vehicle wing shape, and an unmanned aerial vehicle size, the dynamic feature includes at least one of an unmanned aerial vehicle flight speed, an unmanned aerial vehicle flight acceleration, and an unmanned aerial vehicle flight direction, and the unmanned aerial vehicle temperature distribution is obtained by performing extraction on a heat peak area of an unmanned aerial vehicle target using an infrared image processing technology, performing calculation on the heat peak area after the extraction, and determining corresponding temperature gradient parameters and distribution range parameters. The unmanned aerial vehicle recognition and result correction module is configured to input the infrared signal into the unmanned aerial vehicle recognition model to perform unmanned aerial vehicle recognition operation, determine a corresponding unmanned aerial vehicle recognition result, receive a real environment temperature, determine a corresponding infrared temperature measurement deviation result according to the real environment temperature and a preset environment compensation algorithm, perform correction operation on the unmanned aerial vehicle recognition result according to the infrared temperature measurement deviation result, determine a correction result of the unmanned aerial vehicle recognition result, and output the correction result of the unmanned aerial vehicle recognition result to a display interface for display.

6. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the steps of the method for recognizing an unmanned aerial vehicle based on infrared temperature measurement according to any one of claims 1 to 4 when executing the program.

7. A computer readable storage medium having stored thereon a computer program, characterized in that The computer program implements the steps of the method for recognizing an unmanned aerial vehicle based on infrared temperature measurement according to any one of claims 1 to 4 when executed by the processor.

Images