Wheel traffic target classification and identification method, system, device and medium based on micro-motion information

Abstract

The application discloses a wheeled traffic target classification and identification method, system, device and medium based on micro-motion information, and the method comprises the following steps: acquiring a micro-Doppler echo signal of a wheeled target; performing a short-time Fourier time-frequency transformation on the micro-Doppler echo signal to obtain a time-frequency distribution of the micro-Doppler echo signal; calculating an instantaneous frequency of the micro-Doppler echo signal, performing an EMD transformation on the instantaneous frequency, obtaining a plurality of modal components, and selecting a single-frequency component from the plurality of modal components to obtain a wheel rotation period; according to the time-frequency distribution of the micro-Doppler echo signal, the number of flicker points in the wheel rotation period is obtained, and the corresponding maximum speed in the time-frequency distribution is also obtained; and the data set composed of the number of flicker points in the wheel rotation period, the maximum speed and the corresponding vehicle type is input into a support vector machine to classify and identify the vehicle. The application can finely classify and identify different wheeled vehicles.

Description

Technical Field

[0001] This invention belongs to the field of target classification and recognition, specifically relating to a method, system, device, and medium for classifying and recognizing wheeled traffic targets based on micro-motion information. Background Technology

[0002] Statistics show that my country's motor vehicle ownership has experienced significant growth, reaching 435 million in 2023, with cars dominating at 336 million. According to the latest data, 94 cities nationwide have surpassed one million vehicles, with Chengdu, Beijing, Chongqing, Shanghai, and Suzhou exceeding five million, demonstrating their leading position in economic development and population size. In recent years, with the rapid development of new business models such as food delivery, express delivery, and ride-sharing, the number of non-motorized vehicles has also surged. Often, due to time constraints, motor vehicles speed, and non-motorized vehicles frequently appear in motor vehicle lanes, increasing traffic risks and posing significant challenges to road traffic supervision. Existing road target detection and classification systems primarily target motor vehicles, with limited research on non-motorized vehicle identification. With the development of intelligent transportation and increasing demands for environmental perception, the supervision of road traffic targets is becoming increasingly stringent. Therefore, precise classification and identification of traffic targets such as cars, electric vehicles, and bicycles to facilitate timely and optimal handling is particularly important.

[0003] Currently, road traffic detection equipment mainly consists of video detection equipment, lidar, and millimeter-wave radar. Video detection is a common and traditional method, but the optical sensors in image recognition technology are susceptible to the effects of oblique sunlight, rain, fog, and other weather conditions, leading to poor recognition rates and weak anti-interference performance in adverse weather or when obstacles are present. These problems prevent road detection systems from achieving all-weather, all-time remote monitoring of road targets. Compared to laser and infrared detection devices, radar can detect distant targets day and night, is unaffected by fog, clouds, or rain, and has all-weather, all-time characteristics, as well as a certain degree of penetration capability. LiDAR and millimeter-wave radar are newer detection methods developed in recent years. In the field of intelligent transportation, they are mainly used for ranging and speed measurement, as well as obstacle detection, multi-target tracking, collision avoidance system design, and autonomous driving. However, research on the precise classification and identification of road traffic targets is limited, especially for the fine classification of wheeled vehicles such as cars, electric vehicles, and bicycles.

[0004] The micro-Doppler effect is a unique manifestation of target motion characteristics, reflecting the fine motion features of the target. Current research on the application of the micro-Doppler effect mainly focuses on the classification and identification of wheeled vehicles, tracked vehicles, and pedestrians; no reports have been found on the fine classification and identification of wheeled vehicles. Summary of the Invention

[0005] This invention addresses the shortcomings of existing technologies by providing a method, system, device, and medium for classifying and identifying wheeled traffic targets based on micro-motion information. This method is a refined classification and identification approach for wheeled traffic targets such as cars, electric vehicles, and bicycles, based on radar target micro-motion information. It includes several parts such as target micro-motion feature establishment, target micro-motion feature extraction, and refined classification and identification of wheeled targets. This target classification and identification method is based on the micro-Doppler effect of targets, effectively avoiding the shortcomings of existing technologies that are susceptible to the influence of light, fog, clouds, and rain, resulting in either inability to identify targets or poor recognition rates. Furthermore, it enables refined classification and identification of wheeled traffic targets in all weather conditions and at all times.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for classifying and recognizing wheeled traffic targets based on micro-motion information includes the following steps:

[0008] Step 1: Acquire the micro-Doppler echo signal of the wheeled target;

[0009] Step 2: Perform a short-time Fourier transform on the micro-Doppler echo signal to obtain the time-frequency distribution of the micro-Doppler echo signal;

[0010] Step 3: Calculate the instantaneous frequency of the micro-Doppler echo signal, and perform EMD transformation on the instantaneous frequency to obtain multiple modal components. Select a single frequency component from the multiple modal components to determine the wheel rotation period.

[0011] Step 4: Based on the time-frequency distribution of the micro-Doppler echo signal, determine the number of flash points within the wheel rotation cycle, and simultaneously determine the maximum vehicle speed corresponding to the time-frequency distribution;

[0012] Step 5: Input the dataset consisting of the number of flashing points, the maximum vehicle speed, and the corresponding vehicle type within the wheel rotation cycle into the support vector machine to classify and identify the vehicles.

[0013] To optimize the above technical solution, the specific measures also include:

[0014] Further, in step 3, the calculation of the instantaneous frequency of the micro-Doppler echo signal specifically involves:

[0015] Find the micro-Doppler echo signal s Σ instantaneous phase of (t) For instantaneous phase After unwrapping, the differential is calculated to obtain the instantaneous frequency f(t) of the micro-Doppler echo signal, where t represents time. The calculation formula is as follows:

[0016]

[0017] In the formula, unwrap(·) represents unwrapping.

[0018] Further, in step 3, the EMD transformation of the instantaneous frequency specifically involves:

[0019] First, initialize the number of decomposition modes M, and then perform variational mode decomposition on the instantaneous frequency f(t) of the micro-Doppler echo signal to obtain M mode components u. m (t), m=1,2,…M, calculate the energy convergence factor Δ of the modal components at the current decomposition level. M The formula is as follows:

[0020]

[0021] in, Let Δ represent the square of the L2 norm, and set the energy convergence factor threshold ε. M >ε, correct the decomposition mode number M = M + 1, and re-perform variational mode decomposition of the instantaneous frequency of the micro-Doppler echo signal until Δ M If the value is less than or equal to ε, the loop ends, thus obtaining the optimal number of modal components.

[0022] Furthermore, in step 3, the method used to determine the wheel rotation period is the zero-crossing method, specifically:

[0023] For the selected single-frequency component u(t), search for the time point t corresponding to u(t) = 0. i (1≤i≤R), where R represents the number of time points where the single-frequency component u(t) is 0. The wheel rotation period T is calculated using the following formula:

[0024] T = 2mean(t) i+1 -t” i )

[0025] Here, mean(·) represents calculating the average.

[0026] Furthermore, in step 4, the specific steps for determining the number of flashing points within the wheel's rotation cycle are as follows:

[0027] Step 4.1: For the time-frequency distribution TF(t,f), take the cross-section |TF(t,0)| where the frequency f is 0, and search for the time point in the cross-section where |TF(t,0)| reaches its maximum value, and obtain the corresponding time t';

[0028] Step 4.2: Set a threshold α, take the cross-section |TF(t',f)|, and search for the maximum frequency f corresponding to when |TF(t',f)| is greater than α. max The corresponding velocity V of the target is obtained as follows:

[0029]

[0030] where f c is the carrier frequency of the radar, and c is the speed of light;

[0031] Step 4.3: Take the time-frequency distribution section |TF(t, f max )|, search for the value of |TF(t, f max )| ≥ α, and obtain the time corresponding to each scintillation point as t i , (0 < i ≤ S), where S represents the total number of scintillation points in the time-frequency distribution section, and calculate the difference t i+1 - t i ; then the average interval between adjacent scintillation points is:

[0032]

[0033] Step 4.4: Calculate the number of scintillation points N within the period T as:

[0034]

[0035] From this, it is obtained that within one rotation of the wheel, it scintillates N times, and the wheel has N spokes.

[0036] The present invention also proposes a wheeled traffic target classification and recognition system based on micro-motion information, including:

[0037] A Doppler radar for obtaining the micro-Doppler echo signal of the wheeled target;

[0038] A time-frequency transformation module for performing short-time Fourier time-frequency transformation on the micro-Doppler echo signal to obtain the time-frequency distribution of the micro-Doppler echo signal;

[0039] A first calculation module for calculating the instantaneous frequency of the micro-Doppler echo signal and performing EMD transformation on the instantaneous frequency to obtain multiple modal components, and selecting a single-frequency component from the multiple modal components to obtain the wheel rotation period;

[0040] A second calculation module for obtaining the number of scintillation points within the wheel rotation period according to the time-frequency distribution of the micro-Doppler echo signal, and simultaneously obtaining the maximum vehicle speed corresponding in the time-frequency distribution;

[0041] A classification and recognition module for sending the data set composed of the number of scintillation points within the wheel rotation period, the maximum vehicle speed, and the corresponding vehicle type into a support vector machine to classify and recognize the vehicle.

[0042] The present invention also proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the wheeled traffic target classification and recognition method based on micro-motion information as described above.

[0043] The present invention also proposes a computer-readable storage medium storing a computer program that enables a computer to execute the wheeled traffic target classification and recognition method based on micro-motion information as described above.

[0044] The beneficial effects of this invention are:

[0045] 1. Since the target classification and identification method of the present invention is based on a micro-Doppler target detection dedicated radar, it is not easily affected by weather and light conditions such as clouds, rain, and fog, and can work continuously around the clock.

[0046] 2. Each type of vehicle has a fixed wheel model with a fixed design. For automobiles, wheels are roughly divided into four categories based on the number of spokes. Six-spoke wheels are often optional on some American and off-road models, while five-spoke, seven-spoke, and ten-spoke (and more than ten-spoke wheels also fall into this category) wheels cover most mainstream models. Bicycle wheels typically have 16-48 spokes, usually an even number; road bikes often use 20 spokes in the front and 32 or 24 in the rear. Electric bikes now almost all use a one-piece wheel design, essentially a rotating point target. Different wheel designs result in different micro-Doppler effects, leading to different numbers of flickering points within one rotation. Different types of vehicles also exhibit different maximum speeds during operation. This invention uses these two characteristics to finely classify and identify different wheeled vehicles.

[0047] 3. This invention is simple to implement, requires little computation, and can be used for real-time classification and identification of wheeled vehicles in intelligent transportation. Attached Figure Description

[0048] Figure 1 This is a flowchart of the wheeled traffic target classification and recognition method based on micro-motion information proposed in this invention;

[0049] Figure 2 shows the micro-Doppler time-frequency distribution and single-frequency signal diagram when the number of spokes on a car wheel is 5. Figure 2(a) is the single-frequency signal diagram and Figure 2(b) is the micro-Doppler time-frequency distribution diagram. Detailed Implementation

[0050] The invention will now be described in further detail with reference to the accompanying drawings.

[0051] Example 1

[0052] This invention proposes a method for classifying and recognizing wheeled traffic targets based on micro-motion information. The flowchart of this method is as follows: Figure 1 As shown, it includes the following steps:

[0053] Step 1: Obtain the micro-Doppler echo signal s of the wheeled target using a Doppler radar installed on the road. Σ (t);

[0054] Step 2: Analyze the micro-Doppler echo signal s Σ Perform a short-time Fourier transform (t) to obtain the time-frequency distribution TF(t,f) of the micro-Doppler echo signal; the micro-Doppler time-frequency distribution diagram is shown in Figure 2(b).

[0055] Step 3: Calculate the micro-Doppler echo signal s Σ The instantaneous frequency f(t) of (t) is obtained, and EMD transformation is performed on the instantaneous frequency to obtain multiple modal components. From the multiple modal components, a single frequency component s is selected. s (t), the single-frequency signal diagram is shown in Figure 2(a), to determine the wheel rotation period T. Specifically, it includes:

[0056] Step 3.1: Determine the micro-Doppler echo signal s Σ instantaneous phase of (t) For instantaneous phase After unwrapping, the differential is calculated to obtain the instantaneous frequency f(t) of the micro-Doppler echo signal, where t represents time. The calculation formula is as follows:

[0057]

[0058] In the formula, unwrap(·) represents unwrapping.

[0059] Step 3.2: Decompose the instantaneous frequency f(t) of the micro-Doppler signal using empirical mode decomposition, and eliminate local tilt using the adaptive noise complete set empirical mode decomposition method. First, initialize the number of decomposition modes M, and perform variational mode decomposition on the instantaneous frequency f(t) of the micro-Doppler echo signal to obtain M mode components u. m (t), m=1,2,…M, calculate the energy convergence factor Δ of the modal components at the current decomposition level. M The formula is as follows:

[0060]

[0061] in, Let Δ represent the square of the L2 norm, and set the energy convergence factor threshold ε. M >ε, correct the decomposition mode number M = M + 1, and re-perform variational mode decomposition of the instantaneous frequency of the micro-Doppler echo signal until Δ M If the value is less than or equal to ε, the loop ends, thus obtaining the optimal number of modal components.

[0062] Step 3.3: Obtain the wheel rotation period using the zero-crossing method, specifically as follows:

[0063] For the selected single-frequency component u(t), search for the time point t” corresponding to u(t)=0 i , (1≤i≤R), where R represents the number of time points when the value of the single-frequency component u(t) is 0. Use the following formula to obtain the wheel rotation period T:

[0064] T = 2mean(t” i+1 -t” i )

[0065] where mean(·) represents taking the average.

[0066] Step 4: According to the time-frequency distribution TF(t,f) of the micro-Doppler echo signal, obtain the number of scintillation points within the wheel rotation period T, and at the same time obtain the corresponding maximum vehicle speed within the time-frequency distribution TF(t,f). Obtaining the number of scintillation points within the wheel rotation period specifically includes:

[0067] Step 4.1: For the time-frequency distribution TF(t,f), take the section |TF(t,0)| at frequency f = 0, and search for the time point that makes |TF(t,0)| reach the maximum value in the section, and obtain the corresponding time as t';

[0068] Step 4.2: Set the threshold α, take the section |TF(t',f)|, and search for the maximum frequency f corresponding to |TF(t',f)| > α max , and obtain the corresponding speed V of the target as:

[0069]

[0070] where f c is the carrier frequency of the radar, and c is the speed of light;

[0071] Step 4.3: Take the time-frequency distribution section |TF(t,f max )|, and search for the value of |TF(t,f max )|≥α, and obtain the time corresponding to each scintillation point as t i , (0 < i ≤ S), where S represents the total number of scintillation points in the time-frequency distribution section, and calculate the difference between the times t i+1 -t i ; then the average interval between adjacent scintillation points is:

[0072]

[0073] Step 4.4: Calculate the number of scintillation points N within the period T as:

[0074]

[0075] Therefore, if the wheel flashes N times during one revolution, then the wheel has N spokes.

[0076] Step 5: Input the dataset consisting of the number of flashing points, the maximum vehicle speed, and the corresponding vehicle type within the wheel rotation cycle (as shown in Table 1, which is a partial dataset) into a Support Vector Machine (SVN) to classify and identify the vehicles.

[0077] Table 1. Characteristics of Some Wheeled Transportation Targets

[0078]

[0079] Example 2

[0080] The present invention also proposes a wheeled traffic target classification and recognition system based on micro-motion information, corresponding to the method of Embodiment 1, comprising:

[0081] Doppler radar is used to acquire micro-Doppler echo signals from wheeled targets.

[0082] The time-frequency transformation module is used to perform short-time Fourier time-frequency transformation on the micro-Doppler echo signal to obtain the time-frequency distribution of the micro-Doppler echo signal;

[0083] The first calculation module is used to calculate the instantaneous frequency of the micro-Doppler echo signal, perform EMD transformation on the instantaneous frequency to obtain multiple modal components, and select a single frequency component from the multiple modal components to obtain the wheel rotation period.

[0084] The second calculation module is used to calculate the number of flash points within the wheel rotation cycle based on the time-frequency distribution of the micro-Doppler echo signal, and at the same time calculate the maximum vehicle speed corresponding to the time-frequency distribution.

[0085] The classification and recognition module is used to feed the dataset consisting of the number of flashing points, the maximum vehicle speed, and the corresponding vehicle type within the wheel rotation cycle into the support vector machine to classify and recognize the vehicles.

[0086] The implementation methods of each module and its function in the system are completely consistent with the steps of the method in Implementation Example 1, so they will not be repeated here.

[0087] Example 3

[0088] The present invention also proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the wheeled traffic target classification and recognition method based on micro-motion information as described in Embodiment 1.

[0089] Example 4

[0090] The present invention also proposes a computer-readable storage medium storing a computer program that causes a computer to execute the wheeled traffic target classification and recognition method based on micro-motion information as described in Embodiment 1.

[0091] In the embodiments disclosed in this application, a computer storage medium may be a tangible medium that may contain or store programs for use by or in conjunction with an instruction execution system, apparatus, or device. The computer storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of computer storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0092] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0093] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should be considered within the scope of protection of the present invention.

Claims

1. A method for classifying and recognizing wheeled traffic targets based on micro-motion information, characterized in that, Includes the following steps: Step 1: Acquire the micro-Doppler echo signal of the wheeled target; Step 2: Perform a short-time Fourier transform on the micro-Doppler echo signal to obtain the time-frequency distribution of the micro-Doppler echo signal; Step 3: Calculate the instantaneous frequency of the micro-Doppler echo signal, and perform EMD transformation on the instantaneous frequency to obtain multiple modal components. Select a single-frequency component from the multiple modal components to determine the wheel rotation period. In Step 3, the method used to determine the wheel rotation period is the zero-crossing method, specifically: For the selected single-frequency component ,search The corresponding time point R represents a single-frequency component. The number of time points where the value of 0 is 0 is used to calculate the wheel rotation period T using the following formula: in, This indicates calculating the average. Step 4: Based on the time-frequency distribution of the micro-Doppler echo signal, determine the number of flashing points within the wheel rotation cycle, and simultaneously determine the maximum vehicle speed corresponding to the time-frequency distribution; Specifically, determining the number of flashing points within the wheel rotation cycle in Step 4 involves: Step 4.1: For time-frequency distribution Take frequency f A cross section with a value of 0 Search within the section to let The time point at which the maximum value is obtained corresponds to the following time: ; Step 4.2: Set the threshold Take the cut surface ,search Greater than The maximum frequency corresponding to time To obtain the target's corresponding speed. V for: in, The carrier frequency of the radar. The speed of light; Step 4.3: Take the time-frequency distribution section ,search The value is used to obtain the time corresponding to each flashing point. , S This represents the total number of scintillation points in the time-frequency distribution section, and calculates the time difference between points. The average interval between adjacent flashing points; for: Step 4.4: Calculate the number of flashing points within period T. N for: Therefore, it can be seen that within one revolution of the wheel, there is a total of flashing. N Next, the wheel has N One spoke; Step 5: Input the dataset consisting of the number of flashing points, the maximum vehicle speed, and the corresponding vehicle type within the wheel rotation cycle into the support vector machine to classify and identify the vehicles.

2. The wheeled traffic target classification and recognition method based on micro-motion information as described in claim 1, characterized in that, In step 3, the calculation of the instantaneous frequency of the micro-Doppler echo signal specifically involves: Find the micro-Doppler echo signal instantaneous phase For instantaneous phase After untangling, the differential is calculated to obtain the instantaneous frequency of the micro-Doppler echo signal. t represents time, and the calculation formula is: In the formula, It indicates untangling.

3. The wheeled traffic target classification and recognition method based on micro-motion information as described in claim 1, characterized in that, In step 3, the EMD transformation of the instantaneous frequency specifically involves: First, initialize the number of decomposition modes M, and then calculate the instantaneous frequency of the micro-Doppler echo signal. Variational mode decomposition yields M modal components. , Calculate the energy convergence factor of the modal components at the current decomposition level. The formula is as follows: in, Represents the square of the L2 norm, and sets the energy convergence factor threshold. ,like Correcting the number of decomposed modes The instantaneous frequency of the micro-Doppler echo signal is re-decomposed using variational mode decomposition until... The loop ends, thus obtaining the optimal number of modal components.

4. A wheeled traffic target classification and recognition system based on micro-motion information, characterized in that, include: Doppler radar is used to acquire micro-Doppler echo signals from wheeled targets. The time-frequency transformation module is used to perform short-time Fourier time-frequency transformation on the micro-Doppler echo signal to obtain the time-frequency distribution of the micro-Doppler echo signal; The first calculation module is used to calculate the instantaneous frequency of the micro-Doppler echo signal, perform EMD transformation on the instantaneous frequency to obtain multiple modal components, and select a single-frequency component from the multiple modal components to determine the wheel rotation period; the method used to determine the wheel rotation period is the zero-crossing method, specifically: For the selected single-frequency component ,search The corresponding time point R represents a single-frequency component. The number of time points where the value of 0 is 0 is used to calculate the wheel rotation period T using the following formula: in, This indicates calculating the average. The second calculation module is used to determine the number of flashing points within the wheel rotation cycle based on the time-frequency distribution of the micro-Doppler echo signal, and simultaneously determine the maximum vehicle speed corresponding to the time-frequency distribution; specifically, determining the number of flashing points within the wheel rotation cycle involves: Step 4.1: For time-frequency distribution Take frequency f A cross section with a value of 0 Search within the section to let The time point at which the maximum value is obtained corresponds to the following time: ; Step 4.2: Set the threshold Take the cut surface ,search Greater than The maximum frequency corresponding to time To obtain the target's corresponding speed. V for: in, The carrier frequency of the radar. The speed of light; Step 4.3: Take the time-frequency distribution section ,search The value is used to obtain the time corresponding to each flashing point. , S This represents the total number of scintillation points in the time-frequency distribution section, and calculates the time difference between points. The average interval between adjacent flashing points; for: Step 4.4: Calculate the number of flashing points within period T. N for: Therefore, it can be seen that within one revolution of the wheel, there is a total of flashing. N Next, the wheel has N One spoke The classification and recognition module is used to feed the dataset consisting of the number of flashing points, the maximum vehicle speed, and the corresponding vehicle type within the wheel rotation cycle into the support vector machine to classify and recognize the vehicles.

5. An electronic device, characterized in that, include: The memory, the processor, and the computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the wheeled traffic target classification and recognition method based on micro-motion information as described in any one of claims 1-3.

6. A computer-readable storage medium, characterized in that: The computer program stores a computer program that causes the computer to execute the wheeled traffic target classification and recognition method based on micro-motion information as described in any one of claims 1-3.

Images