Airport runway adaptive background elimination system and method

Abstract

The invention relates to the technical field of airport monitoring, and discloses an airport runway adaptive background elimination system and method, and the system comprises a signal collection module which is used for obtaining a real-time monitoring signal of an airport runway area; the multi-scale analysis module is used for carrying out multi-scale analysis on the signals to extract features of background clutters; the background estimation module is used for performing background estimation on the signals through a deep learning algorithm and generating background signals; the threshold adjusting module is used for dynamically adjusting the detection threshold of the target signal according to the background signal; and the filtering module is used for suppressing background clutters through a self-adaptive sliding filter and enhancing target signals. Through multi-module cooperative signal processing, a target signal can be effectively separated from a complex background environment, so that the target signal is accurately detected under runways made of different materials and changing weather conditions, and the problems of false alarm and missing detection caused by background interference in the prior art are avoided.

Description

Technical Field

[0001] The invention relates to the technical field of airport monitoring, and in particular to an airport runway adaptive background elimination system and method. Background Art

[0002] In the airport runway safety monitoring system, the detection of small targets has always been a challenging technical problem. Existing technologies mainly rely on radar, lidar, optical sensors and other devices to collect signals and perform target detection through traditional signal processing methods. Traditional methods usually use fixed thresholds and simple background modeling techniques, such as adaptive constant false alarm (CFAR) algorithms, to distinguish between target and background signals. These technologies have achieved certain results in some applications, especially in the case of simpler background environments and less signal interference, and can achieve more stable target detection.

[0003] However, in practical applications, especially in complex environments such as airport runways, the adaptability and accuracy of existing technologies are often limited. The complexity of the background environment, such as weather changes, runway surfaces of different materials, and changes in time periods, makes the background clutter signal highly dynamic. Existing signal processing methods, such as static thresholds and traditional background estimation methods, often cannot adaptively adjust when dealing with these changes, making it difficult to distinguish between targets and clutter. In addition, traditional filtering techniques usually use fixed filters and window parameters, which makes the filtering effect less than ideal in complex backgrounds. Although polarization signal processing can improve signal resolution, existing processing methods lack sufficient flexibility and cannot effectively cope with polarization changes in different environments, limiting their potential. Summary of the invention

[0004] In view of the deficiencies of the prior art, the present invention provides an airport runway adaptive background elimination system and method, which solves the problems of inaccurate target signal detection, high false alarm and missed alarm rates in complex background environments in the prior art.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: an airport runway adaptive background elimination system, comprising: Signal acquisition module, used to obtain real-time monitoring signals of the airport runway area; A multi-scale analysis module is used to perform multi-scale analysis on the signal to extract the characteristics of background clutter; Background estimation module, which estimates the background of the signal through deep learning algorithm and generates background signal; A threshold adjustment module, used to dynamically adjust the detection threshold of the target signal according to the background signal; The filtering module suppresses background clutter and enhances target signals through an adaptive sliding filter; Polarization signal processing module, using single polarization, dual polarization or multi-polarization transmission and reception mode, adaptively selects polarization mode according to different runway materials and background environments, and optimizes the separation of target signals and background clutter; The target detection module is used to detect the target signal after optimization processing and output the target alarm signal.

[0006] Preferably, the signal acquisition module includes a radar sensor, an optical sensor or a laser radar, which can provide a real-time echo signal of the airport runway area.

[0007] Preferably, the multi-scale analysis module uses wavelet transform, wherein Haar wavelet or Daubechies wavelet basis function is selected to decompose the signal to extract the characteristics of background clutter at different frequency scales.

[0008] Preferably, the background estimation module performs background estimation on the signal through a convolutional neural network, and the convolutional neural network includes multiple convolutional layers, pooling layers and fully connected layers for optimizing the estimation of the background signal.

[0009] Preferably, the threshold adjustment module adjusts the detection threshold in real time according to the estimated value of the background signal, so that the false alarm rate and the missed alarm rate under different background conditions are maintained at a low level.

[0010] Preferably, the filtering module adopts adaptive sliding filtering technology to dynamically adjust the filter window size and filter coefficient according to the frequency change and intensity change of the signal, and the filter dynamically adjusts the filter window size and weight according to the background intensity and frequency change.

[0011] Preferably, the filtering module combines a machine learning algorithm to train and optimize filter parameters and predict optimal filtering windows and coefficients.

[0012] Preferably, the polarization signal processing module can process the target signal and the clutter signal in the polarization domain, and adopts a polarization echo difference analysis model to improve the resolution of the target signal by the difference in polarization characteristics between the target and the clutter.

[0013] Preferably, the polarization signal processing module can automatically select single polarization, dual polarization or multi-polarization mode according to different background environments, and adaptively select the optimal polarization mode for signal processing according to the current runway material and environmental characteristics.

[0014] The airport runway adaptive background elimination method comprises the following steps: S1. Obtain real-time monitoring signals of the airport runway area; S2, perform multi-scale analysis on the signal to extract the characteristics of background clutter; S3, performing background estimation on the signal and generating background signal through deep learning algorithm; S4, dynamically adjusting the detection threshold of the target signal according to the background signal; S5, suppressing background clutter and enhancing target signals through adaptive sliding filters; S6. Use single-polarization, dual-polarization or multi-polarization transmission and reception modes to optimize the separation of target signals and background clutter, and select the appropriate polarization mode according to the background environment; S7, detecting the target signal after optimization processing and outputting the target alarm signal.

[0015] The present invention provides an airport runway adaptive background elimination system and method, which has the following beneficial effects: 1. The present invention adopts a multi-module collaborative signal processing solution, including signal acquisition, background estimation, threshold adjustment, filtering and polarization signal processing, which can effectively separate the target signal from the complex background environment. Through the optimization of each module, the system can accurately detect the target signal under runways of different materials and changing weather conditions. Compared with the processing method of a single module in the prior art, the present invention can achieve more accurate target detection, especially in a strong background clutter environment, avoiding the false alarm and missed detection problems caused by background interference in traditional technologies.

[0016] 2. The present invention adopts adaptive threshold adjustment and adaptive filter technology, which can dynamically optimize the detection sensitivity according to the real-time environmental conditions. The scheme adjusts the threshold and filter parameters in real time according to the changes in the background signal, so that the target signal detection can maintain high sensitivity and low false alarm rate under different circumstances. Compared with the practice of using fixed thresholds and static filters in the prior art, the present invention significantly improves the accuracy of target detection and solves the problem that fixed thresholds and static filters cannot adapt to complex environments.

[0017] 3. The present invention introduces a polarization signal processing module, which uses the difference between the target signal and the background clutter in the polarization domain to separate the target from the clutter. This method can improve the signal discrimination in complex environments, effectively reduce the interference of background clutter, and ensure the accurate extraction of tiny target signals. Compared with the prior art solution that relies on single spectrum analysis, polarization signal processing greatly enhances the distinguishability of the signal, especially when the background clutter frequency is similar, it can effectively improve the detection rate of the target.

[0018] 4 The present invention uses a background estimation method that combines multi-scale analysis and deep learning to accurately estimate dynamically changing background clutter and optimize target detection. This solution can adapt to background changes and effectively distinguish target signals from clutter signals, solving the problems of low background estimation accuracy and ineffective suppression of background interference in the prior art. Compared with traditional background modeling methods, the deep learning method of the present invention has stronger adaptive capabilities, can handle more complex and changeable background environments, and improves the robustness and stability of the system in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a system module diagram of the present invention; Figure 2 It is a schematic diagram of the method steps of the present invention. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the specification of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] Embodiment 1: Please see attached Figure 1 The embodiment of the present invention provides an airport runway adaptive background removal system, comprising: Signal acquisition module, used to obtain real-time monitoring signals of the airport runway area; A multi-scale analysis module is used to perform multi-scale analysis on the signal to extract the characteristics of background clutter; Background estimation module, which estimates the background of the signal through deep learning algorithm and generates background signal; A threshold adjustment module, used to dynamically adjust the detection threshold of the target signal according to the background signal; The filtering module suppresses background clutter and enhances target signals through an adaptive sliding filter; Polarization signal processing module, using single polarization, dual polarization or multi-polarization transmission and reception mode, adaptively selects polarization mode according to different runway materials and background environments, and optimizes the separation of target signals and background clutter; The target detection module is used to detect the target signal after optimization processing and output the target alarm signal; The signal acquisition module includes a radar sensor, an optical sensor or a laser radar, which can provide real-time echo signals of the airport runway area; The multi-scale analysis module uses wavelet transform, in which Haar wavelet or Daubechies wavelet basis function is selected to decompose the signal to extract the characteristics of background clutter at different frequency scales; The background estimation module estimates the background of the signal through a convolutional neural network. The convolutional neural network includes multiple convolutional layers, pooling layers, and fully connected layers to optimize the estimation of background signals. The threshold adjustment module adjusts the detection threshold in real time according to the estimated value of the background signal, so that the false alarm rate and missed alarm rate under different background conditions are kept at a low level; The filter module uses adaptive sliding filtering technology to dynamically adjust the filter window size and filter coefficient according to the frequency and intensity changes of the signal. The filter dynamically adjusts the filter window size and weight according to the background intensity and frequency changes; The filtering module combines machine learning algorithms to train and optimize filter parameters and predict the optimal filter window and coefficients; The polarization signal processing module can process target signals and clutter signals in the polarization domain, and adopts a polarization echo difference analysis model to improve the resolution of the target signal by using the difference in polarization characteristics between the target and clutter. The polarization signal processing module can automatically select single polarization, dual polarization or multi-polarization mode according to different background environments, and adaptively select the optimal polarization mode for signal processing according to the current runway material and environmental characteristics.

[0022] Specifically, the signal acquisition module of this embodiment acquires reflected signals through radar, optical sensor or laser radar, etc. These signals contain the echo information of runway background clutter and potential targets. The acquired signals are transmitted to the subsequent multi-scale analysis module for further processing.

[0023] Generally speaking, the signal acquisition module can monitor and capture background clutter and target signals in real time in the runway area, ensuring the efficiency and accuracy of the entire system. The quality of signal acquisition directly affects the processing effect of subsequent modules, so the selection of acquisition equipment and its working principle are crucial. According to different application environments, choosing different types of sensors can achieve accurate perception of the runway background.

[0024] In this embodiment, the signal acquisition module uses a variety of sensor devices, including radar, laser radar, and optical sensors. These devices obtain signals in different ways and have different working principles. For example, radar equipment can penetrate the surface material of the runway and is suitable for various weather conditions; laser radar can provide accurate three-dimensional images, which is suitable for high-precision target detection; optical sensors can collect images of the runway area in real time, which is suitable for close-range target monitoring. Through these devices, the signal acquisition module can obtain comprehensive data of the airport runway area in real time.

[0025] In some embodiments, the signal acquisition module selects to use a radar sensor. This sensor can transmit electromagnetic waves in real time in the runway area and receive the reflected echo signal. Radar signals can effectively penetrate runway materials such as cement and asphalt through the characteristics of propagation and reflection to obtain reflected wave information. Specifically, factors such as the transmission frequency range, wavelength and antenna design of the radar equipment will directly affect the propagation characteristics of the signal and the quality of the echo.

[0026] In one possible implementation, the signal acquisition module uses a laser radar for target detection. The laser radar can accurately measure the characteristics of the runway surface and obtain high-precision target data through the principle of laser beam emission and reflection. This method is particularly suitable for high-precision target detection scenarios, especially for detecting small targets (such as birds, objects, etc.), its high resolution and accuracy have significant advantages.

[0027] Optical sensors are mainly used to monitor visual information in the runway area. In some embodiments, optical sensors can work together with other sensors to help identify target objects on the runway by providing high-resolution images. In this case, the signal acquisition module not only relies on the reflected wave information of the radar device, but also comprehensively considers the information of the optical image to further improve the accuracy of target detection.

[0028] In this embodiment, the working process of the signal acquisition module can be described by the following formula: the collected signal is set to y(t), where t is a time variable, and the collected signal includes background clutter and target signal. The signal can be expressed as: y(t)=Signal background (t)+Signal target (t) Among them, Signal background (t) is the background clutter signal, Signal target (t) is the target signal. The signal acquisition module is responsible for capturing these signals with appropriate sampling frequency and sampling accuracy, and providing data support for subsequent modules.

[0029] Specifically, the working principle of the signal acquisition module can be further explained by the following equation. Assuming the acquired signal is y(t), it can be divided into two parts: background clutter signal and target signal, where: y(t)=A background ·f background (t)+A target ·f target (t) Here, A background and A target are the amplitudes of the background signal and the target signal, respectively, and fbackground (t) and f target (t) represents the time variation characteristics of background clutter and target signal respectively. By accurately capturing y(t), the signal acquisition module can accurately distinguish target signal from background clutter in complex background environment, providing clear data for subsequent signal processing.

[0030] In order to enhance the stability of signal acquisition, the device in the signal acquisition module also adopts an adaptive gain control mechanism. According to the real-time acquired signal strength, the system automatically adjusts the sensor gain value to adapt to different environmental conditions. This mechanism can improve the sensitivity of signal acquisition in weak signal conditions, ensuring that the system can collect enough information in time even under weak target signal conditions.

[0031] As an option, the signal acquisition module can also be designed with multi-sensor fusion capabilities. In this implementation, multiple sensor devices (such as radar, lidar, and optical sensors) can work simultaneously and monitor the same target from multiple angles and in multiple ways. The data from these sensors are fused and processed to provide more comprehensive and accurate target information. This multi-sensor fusion technology can significantly improve the reliability of target recognition and play an important role in complex environments.

[0032] Through the above technical implementation, the signal acquisition module provides solid data support for the entire system, enabling subsequent multi-scale analysis, background estimation, target detection and other modules to operate efficiently based on accurate data. The design of the signal acquisition module can adapt to the environmental changes of different airport runways, ensuring the robustness and stability of the system in complex environments.

[0033] The main function of the multi-scale analysis module is to perform detailed analysis on the collected signals and extract the background clutter characteristics at different scales. This process is crucial and determines the quality and accuracy of subsequent signal processing. The task of the multi-scale analysis module is to identify various frequency components from complex signals and perform appropriate analysis based on these components to provide optimized background signals for subsequent modules.

[0034] In this embodiment, a time-frequency analysis method such as wavelet transform is used to decompose the signal. As an effective signal processing tool, wavelet transform can localize the signal in the time domain and frequency domain. This enables it to accurately capture the signal characteristics of different frequency components, thereby extracting the details of the background clutter.

[0035] Generally speaking, wavelet transform can provide higher time-frequency resolution and adapt to the changes of different background signals. For example, background clutter usually shows strong fluctuations in the low frequency band, while the target signal often appears in the higher frequency band. By performing wavelet decomposition on the signal, we can separate the signal components at different frequencies, thereby extracting the target signal from the complex background.

[0036] In one possible implementation, the wavelet transform used uses the Daubechies wavelet basis function. The Daubechies wavelet is suitable for processing complex signals because of its good smoothness and locality. Specifically, the wavelet basis function ψ(t) is used to perform convolution operations on the signal to obtain the decomposition result of the signal. The process can be expressed by the following formula: Among them, y(t) is the original signal, ψ(t) is the wavelet basis function, y scale (t) is the signal decomposition result at different scales. Through this formula, wavelet transform can extract the frequency components of the signal at different scales.

[0037] In this embodiment, the result after wavelet transformation is further adjusted and optimized according to the characteristics of the background signal. Specifically, by reconstructing the signals at different scales, the characteristics of the intensity, frequency distribution and time variation of the background clutter can be identified. In this way, the detailed information of the signal is retained, providing accurate data support for the subsequent target detection and background estimation modules.

[0038] As an option, other time-frequency analysis methods (e.g., short-time Fourier transform) can also be used in this embodiment to process the signal. These methods provide a time-frequency representation of the signal by dividing the signal into multiple time periods and performing frequency analysis respectively. In some embodiments, the short-time Fourier transform may be more suitable for the situation where the background signal has a fast-changing feature.

[0039] Furthermore, the multi-scale analysis module can adaptively adjust the scale of the wavelet transform according to the actual collected signal. This adjustment mechanism allows the system to select the most appropriate scale for analysis based on the different frequency components of the background signal. Specifically, the signal scale adjustment formula is: W(t)=max(W 0 ,γ·ΔA(t)) Among them, W 0 is the minimum window size, γ is the adjustment factor, and ΔA(t) is the change in background signal intensity. This formula ensures that when the signal intensity is large, the window size can adapt to the change in signal, thereby avoiding signal loss or information overload.

[0040] Through the above technical solutions, the multi-scale analysis module can effectively extract useful background information from complex signals. In the specific implementation, the module can not only process data from different types of sensors (such as radar signals, optical signals, etc.), but also adaptively adjust the analysis strategy according to the background characteristics in different environments, thereby providing accurate background estimation information.

[0041] In some embodiments, the module can further combine deep learning algorithms to optimize the selection of wavelet basis functions. Specifically, by training on a large amount of labeled data, the system can learn the most suitable basis functions and dynamically adjust them according to actual conditions. This adaptive optimization capability enables the multi-scale analysis module to cope with various complex background conditions and improves the robustness and accuracy of the system.

[0042] The background estimation module is responsible for further estimating the characteristics of background clutter from the processed signal. The main function of the background estimation module is to accurately estimate the actual changes of the background signal by analyzing the statistical characteristics of the background signal, and use it for subsequent target signal detection and optimization.

[0043] In this embodiment, the background estimation module uses a convolutional neural network (CNN) based on deep learning to estimate the background signal. Specifically, after the signal is processed by the multi-scale analysis module, it has been decomposed into components of different frequency bands. The background estimation module receives these analyzed signal data and estimates the background signal through the CNN model. CNN automatically learns the important features in the signal through its multiple convolutional layers and pooling layers, thereby generating accurate background estimation.

[0044] Generally speaking, background signal estimation is a complex process, especially in the presence of dynamically changing backgrounds and interference. To improve the accuracy of the estimation, the background estimation module not only relies on traditional background modeling techniques, but also combines the advantages of deep learning algorithms to adaptively adjust the estimation process based on actual data. In this way, the background estimation module can cope with challenges in different environments, such as environmental noise and interference under different runway materials, weather conditions, and time changes.

[0045] Specifically, the convolutional neural network structure in the background estimation module contains multiple convolutional layers, pooling layers, and fully connected layers. The convolutional layer is responsible for extracting local features from the input signal, the pooling layer enhances the robustness of the network through dimensionality reduction, and the fully connected layer maps these features to the background estimation value. The network is trained using the back-propagation algorithm with the goal of minimizing the background estimation error.

[0046] In a possible implementation, the input signal of the background estimation module is composed of the result processed by the multi-scale analysis module. Assume that the signal provided by the multi-scale analysis module is yscale (t), where t is time, y scale (t) is the signal extracted by wavelet transform or other time-frequency analysis methods. In the background estimation module, these input signals are processed by convolutional neural network to obtain the estimated value of the background signal The calculation formula is: in, is the background estimation value, and CNN represents the convolutional neural network processing process. Background estimation value It reflects the actual characteristics of the background signal at a given time t. This process automatically adjusts weights and biases through a deep learning algorithm to optimize the estimation of the background signal.

[0047] The accuracy of the background estimation module directly affects the accuracy of subsequent target detection. In some embodiments, by training the convolutional neural network, the system can automatically learn a background model suitable for a specific runway, weather and lighting conditions. Specifically, the network will gradually improve its adaptability to different environmental backgrounds through continuous training and verification. The background estimation module estimates the background components in the signal based on these learned features to ensure that the target can still be accurately detected in complex backgrounds.

[0048] In some embodiments, the background estimation module can also be combined with some traditional background modeling methods (such as adaptive constant false alarm detection method, mean and variance estimation, etc.) for hybrid calculation. These methods provide prior information for the deep learning model by analyzing the statistical characteristics of the signal, further improving the stability and accuracy of the estimation. For example, the following formula can be used to combine traditional methods and deep learning models for background estimation: in, is the background estimate calculated by CNN, is the background estimate obtained by traditional statistical methods, and α is the weighting coefficient. The setting of this weighting coefficient can be adjusted through training to fully utilize the advantages of traditional methods and deep learning.

[0049] As a selection, the background estimation module can also pre-process the input of the network to further improve the accuracy of the estimation. For example, in the pre-processing stage of the signal, the signal can be normalized or denoised. These steps can help remove the interference components in the signal and further improve the quality of the background estimation. In certain embodiments, the background estimation module can even adjust the processing mode of the input signal according to real-time feedback, thereby improving the estimation effect of the background signal.

[0050] Output of the background estimation module It is passed to the threshold adjustment module for subsequent target detection and signal optimization. Through accurate background estimation, the threshold adjustment module can better judge the authenticity of the target signal, thereby improving the detection accuracy of the system.

[0051] In the signals processed by the multi-scale analysis module and the background estimation module, the features of the background clutter have been effectively extracted. Next, the threshold adjustment module is responsible for dynamically adjusting the detection threshold of the target signal based on these processed data. This step directly affects the accuracy and stability of target detection. The accurate setting of the threshold is crucial to correctly distinguish between background signals and target signals.

[0052] In this embodiment, the threshold adjustment module dynamically calculates and adjusts the detection threshold based on the background signal estimation value obtained from the background estimation module and the target signal information collected in real time. In this way, the system can adapt to different background environments to ensure that the target signal can be accurately detected in complex backgrounds.

[0053] In general, the detection sensitivity of the target signal needs to be adjusted according to the changes in the background. If the background clutter is strong, the threshold needs to be appropriately increased to avoid false alarms; conversely, when the background clutter is weak, the threshold should be lowered to ensure that weak target signals are not missed. This adjustment process must respond in a timely manner to cope with the dynamic changes in background clutter.

[0054] Specifically, the threshold adjustment module is based on the background signal estimation value output by the background estimation module. And the currently collected target signal characteristics, calculate the appropriate detection threshold T. The formula is as follows: in, is the estimated value of the background signal output by the background estimation module, and α and β are adjustment coefficients. By dynamically adjusting α and β, the sensitivity of the threshold can be controlled to adapt it to different background conditions. Generally, α and β are optimized through training or real-time feedback to achieve the best detection effect.

[0055] As an option, the threshold adjustment module can also dynamically adjust the threshold based on the characteristics of the target signal. For example, if the target signal is strong, the system can lower the threshold to allow the stronger signal to pass. When the target signal is weak, the system will increase the threshold to ensure that the background signal will not trigger falsely. At this time, the system not only relies on the estimation of the background signal, but also further optimizes the detection threshold based on the characteristics of the target signal itself.

[0056] In a possible implementation, the threshold adjustment module also introduces a historical feedback mechanism. Specifically, the system can continuously adjust α and β based on historical detection results (such as false alarm rate, missed alarm rate, etc.), so that the threshold can be optimized as the environment changes during operation. This feedback mechanism can be described by the following formula: α(t)=α(t-1)+γ·error(t) β(t)=β(t-1)+δ·error(t) Where error(t) is the error value at the current moment, which is usually calculated by comparing the difference between the current detection result and the actual situation. γ and δ are adjustment coefficients used to control the speed of feedback. Through this feedback mechanism, the threshold adjustment module can achieve adaptive adjustment to adapt to the changing environment and signal characteristics.

[0057] Specifically, during the operation of the system, when the actual detection result of the target signal differs from the expected result, the error will be calculated and fed back to the threshold adjustment module. The threshold adjustment module automatically adjusts α and β according to the feedback signal, so that the sensitivity of the next detection is optimized. In this way, the system can continuously adjust and optimize the sensitivity of target detection, thereby reducing the risk of false alarms and missed detections.

[0058] In some embodiments, the threshold adjustment module can also perform local optimization according to the specific requirements of the detection area. For example, for a more complex runway area, the system can process the background signals of different areas separately and set different thresholds for each area. Through local optimization, the system can more accurately adapt to the target detection requirements in different environments.

[0059] In some embodiments, in order to further improve the accuracy of threshold adjustment, the system can combine machine learning algorithms to dynamically adjust the threshold. For example, the threshold can be learned and adjusted in real time by support vector machine (SVM) or reinforcement learning method. In this way, the threshold adjustment module can not only respond to changes in background signals, but also adjust its parameters through learning in long-term operation to adapt to the changing environment and target signal characteristics.

[0060] The filtering module is responsible for further optimizing the signal, removing unnecessary background clutter and enhancing the identifiability of the target signal. The core task of the filtering module is to accurately adjust the filtering parameters under different background conditions through adaptive filtering technology to achieve the best clutter suppression effect. This process is crucial for accurately detecting small targets, especially in complex background environments.

[0061] In this embodiment, the filter module adopts an adaptive sliding filtering method, which ensures that clutter can be effectively suppressed and the target signal can be highlighted under various background conditions by dynamically adjusting the window size and filter coefficient of the filter. This module receives the signal output from the threshold adjustment module, and after further processing, passes it to the subsequent target detection module. The filter module not only needs to eliminate background noise, but also needs to maintain the integrity of the target signal to avoid distortion of the target signal.

[0062] In general, the filter window size and coefficients are dynamically adjusted based on the current background signal changes. Specifically, in areas with strong background clutter, the filter will increase the window size to cover a larger signal range; while in areas with a more stable background or weaker signals, the filter will reduce the window size to improve the resolution of the target signal. The size and shape of the window are dynamically optimized through real-time analysis of the background strength and frequency content.

[0063] Specifically, the filtering module in this embodiment realizes dynamic adjustment of filter parameters through the following formula: Among them, y filtered (t) is the signal after filtering, w k is the weight of the filter, y(tk) is the value of the original signal at time tk, and M is the half width of the filter window. The weight of the filter w k It is dynamically calculated based on the frequency and intensity changes of the signal. When the background signal is strong, the weight will increase to reduce the loss of the signal; when the target signal is weak, the weight will decrease to avoid suppressing the target signal too much.

[0064] Alternatively, the filter weights w k Adaptive adjustment can be performed based on the frequency characteristics of the background signal. For example, the frequency characteristics of the signal can be obtained through methods such as wavelet transform, thereby providing a basis for the parameter optimization of the filter. Specifically, the weight w k It can be expressed as: w k (t) = f(Δf(t), ΔA(t)) Among them, Δf(t) is the frequency change of the background signal, ΔA(t) is the change of signal strength, and f(.) is a mapping function that maps these change parameters to filter weights. This adaptive mechanism can dynamically adjust according to the different characteristics of the background signal, thereby maximizing the filtering effect of the signal.

[0065] In one possible implementation, the filter module also combines machine learning methods to further optimize the parameters of the filter. Specifically, by training the model, the system can learn the filter parameters that best suit the current background based on historical signal data. This method can improve the intelligence level of the filter, so that the filter can not only adjust according to the changes in the real-time signal, but also be optimized based on the existing data. The input of the machine learning model includes the frequency, intensity, waveform and other characteristics of the background signal, and the output is the optimal parameters of the filter. This method can significantly improve the processing efficiency and accuracy of the filter module.

[0066] In some embodiments, the filtering module can also use different filtering strategies for different signal sources (such as radar signals, lidar signals, or optical sensor signals). For example, radar signals usually have strong low-frequency components, while optical sensor signals may contain higher-frequency target information. In this case, the filter can automatically select the most suitable filtering strategy based on the characteristics of the input signal. Specifically, when the frequency characteristic of the input signal is low frequency, the filter may use a low-pass filter to suppress background clutter; when the frequency of the input signal is high frequency, the filter may use a high-pass filter to highlight the target signal.

[0067] In addition, the filter's adaptive ability is not limited to adjustment in the time domain, but can also be optimized in the frequency domain. Specifically, the filter module can identify the frequency components in the signal through spectrum analysis methods and adjust the filter parameters according to the spectrum characteristics of the background signal. This frequency domain optimization method can more accurately process background clutter in different frequency bands and avoid unnecessary weakening of the target signal during the filtering process.

[0068] In some embodiments, the filter module also incorporates a sliding window technique. The sliding window technique can dynamically adjust the size and position of the window when processing a signal. Specifically, the sliding window adjusts the size of the window according to the local characteristics of the signal (such as the strength of the signal, frequency changes, etc.). This flexibility allows the filter module to exhibit strong adaptability in a dynamic signal environment.

[0069] The polarization signal processing module is responsible for further improving the resolution and accuracy of the target signal. This module uses the polarization scattering characteristics to separate the target signal from the background signal based on the difference between the target signal and the background clutter in the polarization domain. Polarization signal processing is particularly important in complex environments, and can effectively overcome the interference of background clutter and ensure that weak targets can be correctly detected in complex backgrounds.

[0070] In this embodiment, the polarization signal processing module uses single polarization transmission and dual polarization reception to extract the target signal. In this module, the electromagnetic waves emitted by the system interact with the target object and clutter in a specific polarization mode to generate reflected signals. By analyzing the polarization characteristics of these reflected signals, the target signal and background clutter can be effectively distinguished.

[0071] Specifically, when a single-polarized electromagnetic wave is transmitted, the signal will propagate in a specific polarization direction. When encountering a target or clutter, the polarization characteristics of the reflected wave will change. The dual-polarized receiver can receive two reflected waves in perpendicular directions. By comparing the difference in polarization characteristics of the two reflected waves, the polarization signal processing module can effectively separate the target signal from the clutter.

[0072] In general, the polarization signal processing module plays a key role in the system, especially in the case of strong background clutter. Since clutter and target signals usually have different scattering characteristics in the polarization domain, the polarization signal processing module can highlight the target signal by analyzing such differences, thereby greatly improving the accuracy and reliability of target detection.

[0073] In a possible implementation, the reflected echo signal s target (f) and background clutter signal s clutter (f) represent the polarization echo of target signal and clutter signal respectively. The expressions of target signal and clutter signal are as follows: Among them A target and A clutter Represent the amplitude of target signal and clutter signal respectively, θ is the incident angle φ target (f) and φ clutter (f) is the phase information of the target signal and the clutter signal. The difference in amplitude and phase characteristics of the polarization echo signal makes the target and clutter have significant differences in performance in the polarization domain. The polarization signal processing module separates the target signal by analyzing these differences.

[0074] As an option, the polarization signal processing module can also adopt a multi-polarization transmission and reception mode. This method further improves the resolution of targets and clutter by simultaneously transmitting and receiving signals in multiple different polarization directions. In some embodiments, the system can adaptively select the optimal polarization mode according to different target characteristics and background environments. Specifically, when the background clutter changes greatly, the system will select a high-sensitivity multi-polarization reception mode to enhance the extraction of target signals.

[0075] In some embodiments, the polarization signal processing module combines the polarization echo difference analysis model to separate the target signal and the clutter signal. Specifically, by modeling the echo difference between the target signal and the clutter signal in different polarization modes, the system can determine whether the signal belongs to the target based on the difference in polarization characteristics between the target signal and the clutter signal. This process can be described by the following signal-to-noise ratio formula: Among them, SNR polarized is the signal-to-noise ratio calculated based on polarization characteristics, s target and clutter Represent the target signal and clutter signal respectively, and ∥.∥ represents the energy or strength of the signal. By calculating the polarization difference of the signal, the system can evaluate the distinction between the target signal and the clutter signal and further optimize the target detection process.

[0076] In practical applications, the polarization signal processing module not only relies on the polarization information of the reflected wave, but also combines other characteristics of the target (such as frequency, amplitude, etc.) for comprehensive judgment. For example, when there is a large difference between the reflected wave of the target and the background clutter in the polarization domain, the system will automatically increase the detection sensitivity of the signal. On the other hand, when the polarization characteristics of the clutter and target signals are similar, the system will enhance the target discrimination by improving the signal-to-noise ratio.

[0077] In one possible implementation, the polarization signal processing module can also adaptively select the polarization mode. Specifically, when the runway background is complex and the clutter frequency is high, the system can select the dual-polarization receiving mode for signal reception to effectively distinguish between the target and the clutter. When the background is relatively stable and the clutter is weak, the single-polarization transmission and reception mode may be more efficient.

[0078] The main function of the target detection module is to use the optimized background estimation value and the processed target signal to determine whether there is a target in the signal, and then output a target alarm signal. The target detection module ensures that the target signal can be accurately detected under various background conditions.

[0079] In this embodiment, the target detection module receives the optimized signal from the polarization signal processing module and detects the target in combination with the background signal estimation value provided by the background estimation module. First, the target detection module determines whether the signal exceeds the background noise level based on the adjusted threshold value to determine whether there is a target signal. If the signal strength exceeds the background noise, the system will further analyze the characteristics of the signal to confirm whether it meets the target characteristics. This process is judged at multiple levels through methods such as matching with known target characteristics and polarization characteristic analysis.

[0080] In general, the detection of target signals will be interfered by background clutter and noise. Therefore, the target detection module does not rely solely on a single signal strength threshold, but instead makes judgments based on a combination of multiple features. These features include the signal's amplitude, frequency characteristics, polarization characteristics, and time variation. Through multiple analyses of these features, the target detection module can accurately distinguish between background clutter and true target signals.

[0081] Specifically, the working principle of the target detection module can be expressed by the following formula: in, represents the detection result of the target signal, is the signal after filtering, is the estimated value of the background signal, and T is the detection threshold.

[0082] In a typical embodiment, the detection of the target signal will first perform a preliminary screening of the signal based on the background signal estimate and the signal strength. Specifically, if the signal strength exceeds the combination of the background estimate and the preset threshold, the system will consider the signal to be a target signal. Then, the target signal module will further analyze the frequency, polarization characteristics, etc. of the signal to ensure that it meets the characteristics of the target signal.

[0083] As an option, the target detection module can also introduce machine learning algorithms, especially classification models based on supervised learning, during the signal feature analysis process. Specifically, the system can train a classification model (such as a support vector machine, decision tree, etc.) using the target signal features and non-target signal features in historical data for training, so that the system can automatically determine whether there is a target signal based on the real-time collected signal.

[0084] In one possible implementation, the target detection module identifies the target by comparing the difference between the target signal and the background signal in the polarization domain. Specifically, the signal provided by the polarization signal processing module is optimized to further enhance the discrimination between the target signal and the background clutter through polarization difference analysis. In this case, the polarization characteristics of the target signal and the background clutter have significant differences in scattering characteristics, so that the target signal can be effectively extracted. The target detection module further confirms whether the signal belongs to the target by matching the difference in polarization signals.

[0085] In addition, the target detection module can also combine time series analysis technology to capture the time domain characteristics of the target signal. Specifically, the target signal usually shows a certain time continuity, while the background clutter is usually random and has large time domain fluctuations. The target detection module further verifies whether the signal meets the time domain characteristics of the target by analyzing the consistency of the signal in time.

[0086] In some embodiments, the target detection module can also adaptively adjust the sensitivity of target detection. For example, in a more complex runway background, the system may increase the detection sensitivity to ensure that no target is missed; while in an environment with a more stable background and less clutter, the system can reduce the sensitivity to reduce false alarms. Specifically, the target detection module can dynamically adjust the detection strategy according to changes in the background environment through continuous learning and optimization.

[0087] Embodiment 2: Please see attached Figure 2 , the airport runway adaptive background elimination method includes the following steps: S1. Obtain real-time monitoring signals of the airport runway area; S2, perform multi-scale analysis on the signal to extract the characteristics of background clutter; S3, performing background estimation on the signal and generating background signal through deep learning algorithm; S4, dynamically adjusting the detection threshold of the target signal according to the background signal; S5, suppressing background clutter and enhancing target signals through adaptive sliding filters; S6. Use single-polarization, dual-polarization or multi-polarization transmission and reception modes to optimize the separation of target signals and background clutter, and select the appropriate polarization mode according to the background environment; S7, detecting the target signal after optimization processing and outputting the target alarm signal.

[0088] Specifically, the steps of the method of the present invention include signal acquisition, signal processing, background estimation, threshold adjustment, filtering, polarization signal processing, and target detection. First, in the signal acquisition step, the system collects signals from the airport runway area through devices such as radar, lidar, or optical sensors. These devices can obtain reflected signals containing background clutter and potential targets in real time. The quality of signal acquisition directly affects the effects of subsequent steps, so the selection of equipment and the design of signal acquisition strategies are very critical. For runways in complex environments, sensors need to have sufficient sensitivity and accuracy to ensure that target signals can be captured in a timely manner.

[0089] After signal acquisition, the signal enters the multi-scale analysis module. This module uses time-frequency analysis methods such as wavelet transform to decompose the signal and extract the background clutter features of different frequency bands in the signal. Through multi-scale analysis, the system can separate the low-frequency background clutter and high-frequency target signal components in the signal, ensuring that potential targets in the signal can be effectively distinguished from the background noise. Wavelet transform provides multi-level analysis of the signal in the time domain and frequency domain, so that the signal features of different frequency bands can be clearly extracted and processed.

[0090] Next, the processed signal is transmitted to the background estimation module. At this stage, the system estimates the background clutter in the signal through deep learning algorithms (such as convolutional neural networks). The background estimation module generates an accurate background model by learning and analyzing the statistical characteristics of the signal. This background model provides reliable reference data for subsequent target detection and threshold adjustment. The output of the background estimation module, that is, the estimated value of the background signal, will then be used for subsequent signal discrimination and processing.

[0091] The background estimate is passed to the threshold adjustment module, which dynamically calculates the detection threshold based on the estimated background signal. In environments with strong background signals, the threshold is appropriately increased to avoid false alarms; when the background is relatively stable, the threshold is reduced to increase the sensitivity of target detection. Through this flexible threshold adjustment, the system can adaptively optimize the detection sensitivity of target signals under different environmental conditions, ensuring that target signals are not missed while reducing the occurrence of false alarms.

[0092] After the threshold is adjusted, the signal enters the filtering module. In this step, the filtering module uses adaptive sliding filtering technology to dynamically adjust the size and coefficients of the filtering window according to the frequency and intensity changes of the signal. The filtering module reduces the impact of background clutter and enhances the clarity of the target signal, ensuring that the subsequent target detection process can be carried out with the best signal quality. The parameters of the filter are continuously optimized according to the real-time characteristics of the signal, and can adapt to the background changes and the different characteristics of the target signal.

[0093] The signal then passes through the polarization signal processing module. This module uses the difference in polarization characteristics between the target signal and the clutter to further optimize the signal. Through single-polarization transmission and dual-polarization reception modes, the polarization signal processing module can effectively distinguish the target signal from the background clutter. In the multi-polarization signal mode, the system can also automatically select the most appropriate polarization mode to further improve the detection accuracy of the target signal.

[0094] Finally, the filtered and polarized signal enters the target detection module. In this module, the system accurately detects the target signal based on the background estimate and the filtered signal. If the signal exceeds the set threshold and its characteristics meet the expectations of the target signal, the system will confirm that the signal is a target signal and trigger an alarm. The target detection module can effectively identify the target signal in the complex runway background and avoid the interference of background clutter.

[0095] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. Airport runway adaptive background removal system, characterized by: include: Signal acquisition module, used to obtain real-time monitoring signals of the airport runway area; A multi-scale analysis module is used to perform multi-scale analysis on the signal to extract the characteristics of background clutter; Background estimation module, which estimates the background of the signal through deep learning algorithm and generates background signal; A threshold adjustment module, used to dynamically adjust the detection threshold of the target signal according to the background signal; The filtering module suppresses background clutter and enhances target signals through an adaptive sliding filter; Polarization signal processing module, using single polarization, dual polarization or multi-polarization transmission and reception mode, adaptively selects polarization mode according to different runway materials and background environments, and optimizes the separation of target signals and background clutter; The target detection module is used to detect the target signal after optimization processing and output the target alarm signal.

2. The airport runway adaptive background removal system according to claim 1, characterized in that: The signal acquisition module includes a radar sensor, an optical sensor or a laser radar, which can provide a real-time echo signal of the airport runway area.

3. The airport runway adaptive background removal system according to claim 1, characterized in that: The multi-scale analysis module uses wavelet transform, wherein Haar wavelet or Daubechies wavelet basis function is selected to decompose the signal to extract the characteristics of background clutter at different frequency scales.

4. The airport runway adaptive background removal system according to claim 1, characterized in that: The background estimation module performs background estimation on the signal through a convolutional neural network, and the convolutional neural network includes multiple convolutional layers, pooling layers and fully connected layers, which are used to optimize the estimation of the background signal.

5. The airport runway adaptive background removal system according to claim 1, characterized in that: The threshold adjustment module adjusts the detection threshold in real time according to the estimated value of the background signal, so that the false alarm rate and the missed alarm rate under different background conditions are kept at a low level.

6. The airport runway adaptive background removal system according to claim 1, characterized in that: The filter module adopts adaptive sliding filtering technology to dynamically adjust the filter window size and filter coefficient according to the frequency change and intensity change of the signal. The filter dynamically adjusts the filter window size and weight according to the background intensity and frequency change.

7. The airport runway adaptive background removal system according to claim 1, characterized in that: The filtering module combines machine learning algorithms to train and optimize filter parameters and predict optimal filtering windows and coefficients.

8. The airport runway adaptive background removal system according to claim 1, characterized in that: The polarization signal processing module can process target signals and clutter signals in the polarization domain, and adopts a polarization echo difference analysis model to improve the resolution of the target signal through the difference in polarization characteristics between the target and the clutter.

9. The airport runway adaptive background removal system according to claim 1, characterized in that: The polarization signal processing module can automatically select a single polarization, dual polarization or multi-polarization mode according to different background environments, and adaptively select the optimal polarization mode for signal processing according to the current runway material and environmental characteristics.

10. The airport runway adaptive background removal method, according to the airport runway adaptive background removal system according to any one of claims 1 to 9, characterized in that: The following steps are involved: S1. Obtain real-time monitoring signals of the airport runway area; S2, perform multi-scale analysis on the signal to extract the characteristics of background clutter; S3, performing background estimation on the signal and generating background signal through deep learning algorithm; S4, dynamically adjusting the detection threshold of the target signal according to the background signal; S5, suppressing background clutter and enhancing target signals through adaptive sliding filters; S6. Use single-polarization, dual-polarization or multi-polarization transmission and reception modes to optimize the separation of target signals and background clutter, and select the appropriate polarization mode according to the background environment; S7, detecting the target signal after optimization processing and outputting the target alarm signal.

Images