Laser and microwave data receiving and processing system based on real-time perception

Abstract

The invention discloses a laser and microwave data receiving and processing system based on real-time perception, particularly relates to the technical field of traffic control, and comprises a laser data acquisition module, a microwave data acquisition module, a multi-source data fusion processing module and a real-time decision output module. The laser module dynamically matches a target motion state through a self-adaptive scanning frequency control unit, and generates anti-interference laser point cloud data; the microwave module receives the reflection signal and generates microwave imaging data; the fusion module adopts a space-time alignment convolution layer to extract laser motion features, applies a motion compensation attention mechanism to analyze microwave scattering characteristics, and fuses the features through a cross attention mechanism to generate a unified matrix; and the decision module analyzes the matrix through the FPGA and outputs a control instruction or an early warning signal according to a preset threshold value. According to the system, advantage complementation of laser and microwave data is realized, and the sensing precision and the response speed of a dynamic target in a complex environment are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of traffic control technology, and more specifically, to a laser and microwave data receiving and processing system based on real-time sensing. Background Technology

[0002] Current dynamic target perception systems in traffic environments, such as vehicle and pedestrian detection devices, typically employ independently operating laser and microwave sensing technologies. Laser scanning systems generate point cloud data by emitting pulses at a fixed frequency and receiving reflected signals, used for high-precision spatial positioning of traffic targets. Microwave radar systems generate two-dimensional imaging data by receiving reflected waves, utilizing target scattering characteristics to achieve all-weather traffic monitoring. After independent processing of the data from both systems, target information is generated through simple data stitching or weighted fusion. Finally, a central processor outputs traffic control commands or warning signals based on preset rules. This solution possesses basic perception capabilities in static or low-speed target scenarios.

[0003] However, in practical use, it still has some drawbacks, such as poor motion adaptability, fixed scanning frequency leading to missed detection of high-speed moving vehicles or point cloud distortion, and inability to dynamically match the acceleration changes of traffic targets such as vehicles; weak data complementarity, simple fusion strategy ignores the deep correlation between laser spatial details and microwave physical properties in traffic target recognition, resulting in a sharp drop in perception accuracy in complex traffic environments; insufficient real-time performance, serial processing flow and general processor computing power limitations cause the system response latency to exceed the standard, failing to meet the millisecond-level decision-making requirements of traffic systems; and anti-interference defects, laser data is easily polluted by traffic environment noise, microwave data resolution is insufficient, independent processing amplification error accumulation affects the reliability of traffic control. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a laser and microwave data receiving and processing system based on real-time sensing, which solves the problems mentioned in the background art through the following scheme.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a laser and microwave data receiving and processing system based on real-time sensing, comprising: The laser data acquisition module receives laser reflection signals from traffic target areas in real time through an adaptive scanning frequency control unit. The scanning frequency is dynamically adjusted according to the speed of the traffic target, and the adjustment rate changes synchronously with the target's motion state. It also uses pulse code modulation technology to generate anti-interference laser point cloud data. The microwave data acquisition module receives microwave reflection signals from the traffic target area and generates microwave imaging data. The multi-source data fusion processing module, which is communicatively connected to the laser and microwave data acquisition modules, includes a multimodal feature association network. The anti-interference laser point cloud data is extracted in real time by a spatiotemporally aligned convolutional layer to generate a first target feature vector. A second target feature vector is generated by performing scattering analysis on microwave imaging data using a motion-compensated attention mechanism. The first and second target feature vectors are input into a multimodal feature association network, and feature-level complementary fusion is achieved based on the cross-attention mechanism to generate a fused feature matrix. The real-time decision output module parses the fused feature matrix and generates traffic control instructions or early warning signals according to preset traffic decision rules.

[0006] Preferably, the adaptive scanning frequency control unit includes: a laser reflection signal receiving device for receiving laser reflection signals from traffic target areas in real time and an algorithm component based on PID control; the algorithm component dynamically adjusts the scanning frequency according to the speed of the traffic target, and makes the adjustment rate of the scanning frequency change synchronously with the target motion state characterized by acceleration and velocity.

[0007] Preferably, the PID control-based algorithm component includes: a proportional coefficient adjustment unit, an integral coefficient adjustment unit, a derivative coefficient adjustment unit, and a target speed reference value setting unit; the proportional coefficient adjustment unit adjusts the proportional coefficient in response to changes in the target's motion speed, the integral coefficient adjustment unit adjusts the integral coefficient based on the speed deviation, the derivative coefficient adjustment unit predicts the target's motion trend, and limits excessive adjustment of the scanning frequency through damping effect when the target decelerates rapidly or accelerates abruptly, and the target speed reference value setting unit sets the target speed reference value.

[0008] Preferably, the anti-interference laser point cloud data includes: a sampling frequency determination unit, which determines the sampling frequency according to the Nyquist sampling theorem; a signal quantization unit, which quantizes the sampled signal and determines the quantization interval according to the quantization level range and quantization level; an encoding unit, which encodes the quantized signal according to binary encoding rules; and a three-dimensional matrix generation unit, which generates anti-interference laser point cloud data in the form of a three-dimensional matrix, wherein two dimensions represent the spatial planar coordinates of traffic scenes such as roads or intersections, the third dimension represents the time dimension, and the matrix elements store the feature values ​​of the laser reflection intensity at the corresponding spatial location and time point.

[0009] Preferably, the microwave data acquisition module includes: a microwave reflection signal receiving antenna with high gain and narrow beamwidth; a low-noise amplifier to amplify the microwave reflection signal to enhance signal strength; an analog-to-digital converter to convert the amplified signal into a digital signal; a radar cross section calculation unit, which calculates the radar cross section of traffic targets such as vehicles based on the radar cross section principle by measuring the received power and known transmit power, antenna gain, wavelength and target distance parameters; and a microwave imaging data generation unit, which generates microwave imaging data in the form of a two-dimensional matrix through a signal processing algorithm, where the two dimensions represent the coordinate position of the imaging plane and the matrix elements represent the microwave reflection intensity or related quantization value at the corresponding position.

[0010] Preferably, the matrix generation of the microwave imaging data includes: a target displacement calculation unit calculating the target displacement based on the target's motion speed and the microwave signal propagation time; a spatial translation correction unit performing spatial translation correction on the microwave imaging data based on the target displacement; a two-dimensional image generation unit generating a two-dimensional image from the spatially translated data using a back projection algorithm; and a discretization unit discretizing the two-dimensional image into a microwave imaging data matrix.

[0011] Preferably, the first target feature vector includes: a vector generated by real-time feature extraction of anti-interference laser point cloud data in the form of a three-dimensional matrix through a spatiotemporally aligned convolutional layer, wherein the spatiotemporally aligned convolutional layer binds a timestamp from the clock synchronization signal of the laser scanning system to each spatial coordinate point, calculates the motion vector of the target per unit time through the spatial position difference of two consecutive frames of point cloud data, adopts deformable convolutional kernels, and uses multi-scale convolutional kernels to extract the instantaneous motion features, motion trends and spatial global features of the target respectively, and obtains the target by fusing the feature maps generated by multiple convolutional kernels and performing pooling operations.

[0012] Preferably, the second target feature vector includes: a vector generated by performing scattering analysis on microwave imaging data through a motion-compensated attention mechanism, wherein the motion-compensated attention mechanism first calculates the motion distance based on the target's motion speed and microwave signal propagation time, performs spatial translation compensation on the microwave imaging data, then expands the compensated microwave imaging data matrix into a vector in a specific way, calculates the attention weight matrix, and obtains the scattering analysis result through matrix operations, thereby generating a vector carrying quantitative information related to the target's microwave reflection characteristics.

[0013] Preferably, the fusion feature matrix includes: inputting the first target feature vector and the second target feature vector into a multimodal feature association network, and generating a fusion feature matrix by realizing feature-level complementary fusion based on a cross-attention mechanism. The cross-attention mechanism calculates the cross-attention matrix, which is generated by matrix operation with the second target feature vector as the multiplicand.

[0014] Preferably, the real-time decision output module includes: a high-performance field-programmable gate array chip with high-speed parallel processing capability; a component for parsing and fusing feature matrices, a unit for presetting decision rules and parameter thresholds, and a unit for generating traffic control commands such as traffic light control and gate control or warning signals for triggering warning devices based on the conditions satisfied by the feature values; and a multi-source data fusion processing module connected to a high-speed data transmission bus to optimize the parsing and decision algorithms.

[0015] The technical effects and advantages of this invention are as follows: Enhanced dynamic motion adaptability: Improves the system's dynamic adaptability to changes in the motion state of traffic targets such as vehicles, effectively avoiding missed detection of high-speed vehicles and ensuring smooth and safe traffic flow; Deep complementary fusion of multi-source data: Through a deep feature fusion mechanism, the advantages of laser and microwave data are complemented in traffic target perception, thereby improving the perception accuracy in complex traffic environments. Millisecond-level real-time decision-making capability: Utilizing a hardware parallel processing architecture, the system ensures that it meets the millisecond-level real-time decision-making requirements for traffic control such as intersection signal control and hazard warning; Comprehensive enhancement of anti-interference performance: Enhanced system anti-interference performance ensures the reliability of perception data in variable traffic environments such as heavy traffic and environmental clutter, and improves the accuracy of control commands. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the overall structure of the present invention.

[0017] Figure 2 This is a schematic diagram of the multimodal feature fusion structure of the present invention.

[0018] Figure 3 This is a schematic diagram of the real-time decision-making logic structure of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] refer to Figures 1-3The system shown is a laser and microwave data receiving and processing system based on real-time sensing, including: a laser data acquisition module, a microwave data acquisition module, a multi-source data fusion processing module, and a real-time decision output module. The connection method is as follows: the laser data acquisition module and the microwave data acquisition module are respectively connected to the multi-source data fusion processing module, and the multi-source data fusion processing module is connected to the real-time decision output module.

[0021] The laser data acquisition module receives laser reflection signals from traffic target areas in real time through an adaptive scanning frequency control unit. The scanning frequency is dynamically adjusted according to the speed of the traffic target, and the adjustment rate changes synchronously with the target's motion state. It also uses pulse code modulation technology to generate anti-interference laser point cloud data. As a preferred example of feasibility, the adaptive scanning frequency control unit of the laser data acquisition module employs a high-sensitivity laser reflection signal receiving device to ensure that laser reflection signals from traffic target areas can be received in real time and accurately. The scanning frequency is dynamically adjusted based on the speed of the traffic target. This is achieved using a PID control-based algorithm; the target speed is... The scanning frequency is The initial scan frequency is The proportionality coefficient is The integral coefficient is The differential coefficient is Scan frequency adjustment amount The calculation formula is as follows: ,in The target speed reference value is set. For time; Real-time scanning frequency In this way, the scanning frequency can be adjusted in a timely and precise manner according to changes in the speed of traffic targets. The target's motion state is determined by acceleration. and speed To comprehensively characterize; Adjustment rate The synchronous change with the target's motion state is achieved through the following formula: ,in For microwave signal propagation time, and These coefficients are obtained through experimental calibration based on system hardware performance and practical application requirements; when the target object accelerates, the acceleration... As the frequency increases, the adjustment rate of the scanning frequency also increases accordingly, ensuring that the movement changes of the target object can be tracked in a timely manner. When performing pulse code modulation on a laser reflection signal, the sampling frequency is first determined. According to the Nyquist sampling theorem, the sampling frequency... Must meet ,in It is the highest frequency component in the laser reflection signal; in practical applications, it is generally selected. To reserve a safety margin to ensure the accuracy of sampling; After sampling, the signal is quantized, and the quantization level range is set to... Quantization level is Then the quantization interval The quantized signal is encoded according to binary encoding rules to generate anti-interference laser point cloud data.

[0022] The anti-interference laser point cloud data is finally converted into a three-dimensional matrix. The form of representation, in which and This represents the coordinate position on the spatial plane, corresponding to the sampling point position of the laser scan on the plane; Representing the time dimension, it records anti-interference laser point cloud data information at different times; each matrix element The corresponding spatial location is stored. and time point The laser reflection intensity and other related characteristic values ​​at the location.

[0023] It should be further explained that the proportionality coefficient The primary goal is to ensure the system's ability to respond quickly to changes in target velocity. This includes addressing issues like sudden target acceleration errors. When the ratio is increased, the proportional term dominates the rapid adjustment of the scanning frequency, reducing tracking lag. This avoids overshoot oscillations caused by excessively fast response, balancing sensitivity and stability; integral coefficient This is used to eliminate long-term, small-amplitude velocity deviations. A smaller value can effectively suppress the cumulative amplification of environmental noise during the integration process, preventing system instability caused by integral saturation; the differential coefficient... Its core function is to predict the target's motion trend and suppress overshoot. When the target decelerates rapidly or accelerates abruptly, the differential term limits excessive adjustment of the scanning frequency through a damping effect, preventing system oscillation. The value of... for Half of the original value can effectively smooth the response curve and reduce sensitivity to high-frequency noise, complementing hardware filtering measures.

[0024] It should be further explained that the target speed reference value This is because the value falls within the common speed range of dynamic targets targeted by the laser data acquisition module. The intermediate value of the value can cover the normal motion state of most medium and low speed dynamic targets. It matches the performance constraints of laser scanning hardware, avoiding hardware overload due to excessively high values ​​or frequent invalid adjustments due to excessively low values. It also facilitates parameter calibration, reduces the cumulative error of the integral term, and can maintain a reasonable sampling density at the initial scanning frequency, taking into account both data acquisition efficiency and redundancy. It has wide applicability and practical value.

[0025] The microwave data acquisition module receives microwave reflection signals from the traffic target area and generates microwave imaging data. As a preferred feasible example, the microwave data acquisition module is equipped with a microwave reflection signal receiving antenna with high gain and narrow beamwidth to improve the efficiency and accuracy of receiving microwave reflection signals in traffic target areas; The microwave reflected signal is first amplified by a low-noise amplifier to enhance signal strength for easier subsequent processing; the amplified signal is then converted into a digital signal by an analog-to-digital converter, with a sampling precision set to [value missing]. The microwave imaging data is generated using the principle of microwave radar cross-section; the microwave transmission power is... The antenna gain is The microwave wavelength is The target distance is The received power is Then the radar cross section The calculation formula is: ; through measurement and known , , and Parameters such as radar cross section of target object are calculated, and then microwave imaging data is generated based on a series of signal processing algorithms. Microwave imaging data in two-dimensional matrix The form of representation, in which and Represents the coordinate position on the imaging plane; each matrix element Represents the position in the imaging plane The microwave reflection intensity or the quantitative value related to microwave scattering characteristics at a given location reflects the microwave reflection characteristics of the traffic target area at that location.

[0026] It should be further explained that the signal processing algorithm generates microwave imaging data: Based on the speed of the traffic target and microwave signal propagation time Calculate the target displacement: Spatial translation correction is performed on the microwave imaging data based on the target displacement; a two-dimensional image is generated from the motion-compensated data using a back projection algorithm; and the two-dimensional image is discretized into a microwave imaging data matrix. .

[0027] The multi-source data fusion processing module, which is communicatively connected to the laser and microwave data acquisition modules, includes a multimodal feature association network. The anti-interference laser point cloud data is extracted in real time by a spatiotemporally aligned convolutional layer to generate a first target feature vector. A second target feature vector is generated by performing scattering analysis on microwave imaging data using a motion-compensated attention mechanism. The first and second target feature vectors are input into a multimodal feature association network, and feature-level complementary fusion is achieved based on the cross-attention mechanism to generate a fused feature matrix. As a preferred example of feasibility, spatiotemporally aligned convolutional layers are used with a three-dimensional matrix. The anti-interference laser point cloud data, represented in a formal format, is processed as follows: The convolutional layer uses multiple convolutional kernels of different sizes. , ,in The number of convolution kernels, These represent the convolution kernels at... Dimensions in three dimensions; The formula for convolution is: ; in These represent the dimensions of the convolution kernel in the three dimensions. This represents the three-dimensional coordinates on the output feature map after convolution. It is the traversal index of the convolution kernel along a certain dimension. For the first After convolution by several kernels, The feature maps generated by multiple convolutional kernels are fused and then pooled to finally generate the first target feature vector. .

[0028] For a two-dimensional matrix Microwave imaging data, represented in formal form, is first subjected to motion compensation. This is done if the velocity of the target object during microwave signal transmission and reception is known. and the propagation time of microwave signals Then the target's movement distance Spatial translation compensation is applied to the microwave imaging data based on the motion distance; then, scattering analysis is performed using an attention mechanism; the attention weight matrix is... Its elements Calculated using the following formula: ,in and It is a microwave imaging data matrix A vector expanded in a specific way. It is an eigenvector transpose, Let the dimension be the vector. The number of vectors, For index variables, For index variables, For the circular index variable, It is a microwave imaging data matrix A vector expanded in a specific way; Through matrix operations The results obtained after scattering analysis are then used to generate the second target feature vector. Here is the microwave imaging data matrix. Each element is ,in and Determine the position on the imaging plane, where the element represents the quantized value related to the microwave reflection characteristics at the corresponding position; The first target feature vector Second target feature vector Input a multimodal feature association network for feature-level complementary fusion; Calculate the cross-attention matrix Its elements The calculation formula is: ,in, Indicates the first Individual and Related feature vectors, Indicates the first Individual and Related feature vectors, The dimension of the feature vector; Through matrix operations Generate a fusion feature matrix This enables feature-level complementary fusion of anti-interference laser point cloud data and microwave imaging data.

[0029] It should be further explained that the spatiotemporally aligned convolutional layer: binds precise timestamp information to each spatial coordinate point in the 3D matrix of the laser point cloud data. The timestamps are derived from the clock synchronization signal of the laser scanning system; by using the spatial position difference between two consecutive frames of point cloud data, it calculates the target's motion vector (including horizontal displacement) per unit time. Vertical displacement and time interval ), forming motion vectors Design deformable convolution kernels, whose sampling points have original coordinates. Dynamically adjusted according to the formula ,in As a time decay factor, when the target suddenly accelerates, Increase to 1 for rapid compensation of position offset; when the target moves at a constant speed, Reduced to 0.8 to avoid overcorrection; multi-scale convolutional kernels work together, including short-term tracking kernel, trajectory prediction kernel and global alignment kernel, to extract the instantaneous motion features, motion trends and spatial global features of the target respectively.

[0030] It should be further explained that in the multimodal feature fusion process, the second target feature vector is chosen as the multiplicand instead of the first target feature vector. The reason for this is the physical adaptability of the technical architecture and the synergistic optimization of the computational logic: microwave features carry the essential physical properties of the target object, have strong anti-interference capabilities and all-weather stability, and provide a reliable foundation of basic attributes for fusion. While laser features have high-precision spatial details, they are susceptible to environmental interference, leading to data fluctuations. Through the cross-attention mechanism, laser features act as a dynamic query vector to actively focus on key areas, while microwave features act as enhanced key-value vectors, receiving weight allocation guided by the laser in matrix operations. This achieves dynamic correction of the resolution of microwave data with laser spatial precision, while preserving the integrity of the physical properties of microwaves. If the operation is reversed, using laser features as the multiplicand, not only will environmental noise contaminate the fusion results, but the high-dimensionality of laser data will also significantly increase the computational load, compromising the system's real-time performance.

[0031] The real-time decision output module parses the fused feature matrix and generates traffic control instructions or early warning signals according to preset traffic decision rules.

[0032] As a preferred feasible example, the real-time decision output module uses a high-performance field-programmable gate array chip, which has high-speed parallel processing capabilities and can meet real-time requirements. Chip analysis and fusion feature matrix A series of decision-making rules are preset; threshold values ​​for decision parameters are set. etc.; if a certain eigenvalue in the fused feature matrix If the current state is maintained, monitoring will continue; if a certain eigenvalue in the fused feature matrix is... satisfy and (in and If the index is a specific position in the feature matrix, then control commands are generated, such as controlling related equipment to perform adjustment operations; if ( If another specific location is indexed, a warning signal is generated, and an alarm is issued through devices such as audible and visual alarms.

[0033] To ensure that the system response latency meets the real-time perception requirements, a high-speed data transmission bus is used in the hardware design to connect the multi-source data fusion processing module and the real-time decision output module, reducing data transmission latency. The bus bandwidth must meet the real-time transmission requirements of the fused feature matrix. In terms of algorithm optimization, the analysis and decision algorithms are optimized to reduce unnecessary calculation steps and lower the time complexity of the algorithms.

[0034] It needs to be further explained that, When the fused feature matrix is ​​in uniform target motion and without environmental interference, take It can ensure that more than 95% of normal states do not trigger warnings falsely, and the frequency is below the minimum adjustment threshold of the laser adaptive scanning frequency, avoiding invalid responses caused by minor fluctuations; If the spatiotemporal alignment deviation of the laser point cloud exceeds the correction range of the deformable convolution kernel, or if the fluctuation of the microwave imaging scattering characteristics exceeds the error tolerance range of the back projection algorithm, the parameters need to be actively adjusted. The laser point cloud has more than 30% invalid points and the microwave scattering cross section calculation error exceeds 50%, rendering the fused feature matrix worthless for decision-making.

[0035] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other. In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A laser and microwave data receiving and processing system based on real-time sensing, characterized in that, include: The laser data acquisition module receives laser reflection signals from traffic target areas in real time through an adaptive scanning frequency control unit. The scanning frequency is dynamically adjusted according to the speed of the traffic target, and the adjustment rate changes synchronously with the target's motion state. It also uses pulse code modulation technology to generate anti-interference laser point cloud data. The microwave data acquisition module receives microwave reflection signals from the traffic target area and generates microwave imaging data. The multi-source data fusion processing module, which is communicatively connected to the laser and microwave data acquisition modules, includes a multimodal feature association network. The anti-interference laser point cloud data is extracted in real time by a spatiotemporally aligned convolutional layer to generate a first target feature vector. A second target feature vector is generated by performing scattering analysis on microwave imaging data using a motion-compensated attention mechanism. The first and second target feature vectors are input into a multimodal feature association network, and feature-level complementary fusion is achieved based on the cross-attention mechanism to generate a fused feature matrix. The real-time decision output module parses the fused feature matrix and generates traffic control instructions or traffic warning signals.

2. The laser and microwave data receiving and processing system based on real-time sensing according to claim 1, characterized in that, The adaptive scanning frequency control unit includes: a laser reflection signal receiving device for receiving laser reflection signals from traffic target areas in real time and an algorithm component based on PID control; the algorithm component dynamically adjusts the scanning frequency according to the speed of the traffic target, and makes the adjustment rate of the scanning frequency change synchronously with the target motion state characterized by acceleration and velocity.

3. The laser and microwave data receiving and processing system based on real-time sensing according to claim 2, characterized in that, The PID-based control algorithm components include: a proportional coefficient adjustment unit, an integral coefficient adjustment unit, a derivative coefficient adjustment unit, and a target speed reference value setting unit. The proportional coefficient adjustment unit adjusts the proportional coefficient in response to changes in the speed of the traffic target. The integral coefficient adjustment unit adjusts the integral coefficient based on the speed deviation. The derivative coefficient adjustment unit predicts the target's movement trend and limits excessive adjustment of the scanning frequency through damping effect when the target decelerates rapidly or accelerates abruptly. The target speed reference value setting unit sets the target speed reference value.

4. The laser and microwave data receiving and processing system based on real-time sensing according to claim 1, characterized in that, The anti-interference laser point cloud data includes: a sampling frequency determination unit, which determines the sampling frequency according to the Nyquist sampling theorem; a signal quantization unit, which quantizes the sampled signal and determines the quantization interval according to the quantization level range and quantization level; an encoding unit, which encodes the quantized signal according to binary encoding rules; and a three-dimensional matrix generation unit, which generates anti-interference laser point cloud data in the form of a three-dimensional matrix, where two dimensions represent the spatial planar coordinates of traffic scenes such as roads or intersections, and the third dimension represents the time dimension. The matrix elements store the characteristic values ​​of the laser reflection intensity at the corresponding spatial location and time point.

5. The laser and microwave data receiving and processing system based on real-time sensing according to claim 1, characterized in that, The microwave data acquisition module includes: a microwave reflection signal receiving antenna with high gain and narrow beamwidth; a low-noise amplifier to amplify the microwave reflection signal to enhance signal strength; an analog-to-digital converter to convert the amplified signal into a digital signal; a radar cross section calculation unit, which calculates the radar cross section of the target based on the radar cross section principle by measuring the received power and the known transmit power, antenna gain, wavelength, and target distance parameters; and a microwave imaging data generation unit, which generates microwave imaging data in the form of a two-dimensional matrix through signal processing algorithms, where the two dimensions represent the coordinate position of the imaging plane, and the matrix elements represent the microwave reflection intensity or related quantization value at the corresponding position.

6. The laser and microwave data receiving and processing system based on real-time sensing according to claim 1, characterized in that, The matrix generation of the microwave imaging data includes: a target displacement calculation unit calculates the target displacement based on the speed of the traffic target and the propagation time of the microwave signal; a spatial translation correction unit performs spatial translation correction on the microwave imaging data based on the target displacement; a two-dimensional image generation unit generates a two-dimensional image from the spatially translated data using a back projection algorithm; and a discretization unit discretizes the two-dimensional image into a microwave imaging data matrix.

7. The laser and microwave data receiving and processing system based on real-time sensing according to claim 1, characterized in that, The first target feature vector includes: a vector generated by real-time feature extraction of anti-interference laser point cloud data in the form of a three-dimensional matrix through a spatiotemporally aligned convolutional layer. The spatiotemporally aligned convolutional layer binds a timestamp from the clock synchronization signal of the laser scanning system to each spatial coordinate point. The motion vector of the target per unit time is calculated by the spatial position difference of two consecutive frames of point cloud data. Deformable convolutional kernels are used, and multi-scale convolutional kernels are used to extract the instantaneous motion features, motion trends and spatial global features of the target respectively. The feature maps generated by multiple convolutional kernels are fused and then pooled to obtain the target.

8. The laser and microwave data receiving and processing system based on real-time sensing according to claim 1, characterized in that, The second target feature vector includes a vector generated by performing scattering analysis on microwave imaging data through a motion-compensated attention mechanism. The motion-compensated attention mechanism first calculates the movement distance based on the speed of the traffic target and the propagation time of the microwave signal, performs spatial translation compensation on the microwave imaging data, expands the compensated microwave imaging data matrix into a vector in a specific way, calculates the attention weight matrix, and obtains the scattering analysis result through matrix operations, thereby generating a vector carrying quantitative information related to the microwave reflection characteristics of the target.

9. The laser and microwave data receiving and processing system based on real-time sensing according to claim 1, characterized in that, The fusion feature matrix includes: inputting the first target feature vector and the second target feature vector into a multimodal feature association network, and generating a fusion feature matrix by realizing feature-level complementary fusion based on a cross-attention mechanism. The cross-attention mechanism calculates the cross-attention matrix, which is generated by matrix operation with the second target feature vector as the multiplicand.

10. A laser and microwave data receiving and processing system based on real-time sensing according to claim 1, characterized in that, The real-time decision output module includes: a high-performance field-programmable gate array chip with high-speed parallel processing capability; a component for parsing and fusing feature matrices, a unit for preset decision rules and parameter thresholds, and a component for generating traffic control instructions or traffic warning signals based on the conditions satisfied by the feature values; and a multi-source data fusion processing module connected to a high-speed data transmission bus to optimize the parsing and decision algorithms.

Images