Multi-target identification method in complex traffic environment and electronic equipment
By extracting and fusion of multi-frame features of road point cloud data, the multi-object recognition accuracy in complex traffic environments is improved, and the problem of low accuracy of single-frame data recognition in the prior art is solved, and more accurate target recognition is achieved.
Patent Information
- Application Number
- CN202310574042.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-19
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2043-05-19
AI Technical Summary
The prior art has low accuracy in traffic road vehicle identification in complex traffic environments, especially due to the influence of abnormal detection results and the limitations of single-frame point cloud data.
By obtaining the road point cloud data with a specified number of frames, the point cloud features, HRRP geometric features and HRRP power spectrum features of single frame data are extracted, and the point cloud feature set is formed, and the feature sequence of multi-frame data is input to the pre-trained model for target recognition .
It improves the accuracy of multi-object recognition in complex traffic environments, captures more detailed information of target objects, and enhances the recognition ability in crowded environments.
Smart Images

Figure CN119007128A_ABST
Abstract
Description
Background Art
[0002] Complex traffic environment refers to mixed traffic consisting of motor vehicles, non-motor vehicles and pedestrians, which seriously affects road traffic order, reduces road traffic capacity and increases the difficulty of road traffic management.
[0003] The solution of realizing traffic control, real-time monitoring and statistics of the location, speed and number of various types of vehicles on the road based on traffic road vehicle identification, and preventing traffic accidents from the root is an effective way to quickly improve the road traffic environment and enhance traffic safety.
[0004] In the prior art, the method of realizing traffic road vehicle recognition is to obtain the classification result based on the point cloud data generated by a single detection of the target. However, in the actual road target classification task, the classification effect of this method is easily affected by abnormal detection results, and the accuracy of target recognition based on only a single frame of point cloud data is low in a crowded and complex traffic environment. Summary of the invention
[0005] In an exemplary embodiment of the present disclosure, a method and electronic device for multi-target recognition in a complex traffic environment are provided to improve the accuracy of target recognition in a complex traffic environment.
[0006] A first aspect of the present disclosure provides a method for identifying multiple targets in a complex traffic environment, the method comprising:
[0007] At specified intervals, obtaining road point cloud data of a specified number of frames, wherein the road point cloud data is obtained by detecting the target road using a millimeter wave radar;
[0008] For any frame of road point cloud data, project the road point cloud data on the radial distance of the millimeter wave radar to obtain a high-resolution distance signal HRRP of the road point cloud data, and extract point cloud features from the road point cloud data to obtain point cloud features of each of the road point cloud data; and,
[0009] Obtaining a point cloud feature set of the road point cloud data according to the point cloud features, HRRP geometric features, and HRRP power spectrum features of the road point cloud data, wherein the HRRP geometric features and the HRRP power spectrum features are obtained by extracting features from the HRRP of the road point cloud data;
[0010] Obtaining a point cloud feature sequence according to the point cloud feature set of the road point cloud data of the specified number of frames;
[0011] The point cloud feature sequence is input into a pre-trained target classification model to obtain the category of each target object in the road point cloud data of the specified number of frames.
[0012] In this embodiment, the point cloud feature set of a single frame of road point cloud data is determined by first extracting the point cloud features, HRRP geometric features, and HRRP power spectrum features of the single frame of road point cloud data, and then the point cloud feature sets of multiple frames of road point cloud data are combined to obtain a point cloud feature sequence. Target recognition is performed based on the point cloud feature sequence to obtain the category of each target object. Therefore, in the application embodiment, multiple features in a single frame of road point cloud data are extracted to accurately obtain the point cloud feature set of each target object; and a high-dimensional multi-frame point cloud feature sequence is obtained by fusing the point cloud feature set of multiple frames of road point cloud data on a time series. Therefore, more detailed information of each target object in a complex traffic environment can be captured to improve the accuracy of multi-target recognition in a complex traffic environment.
[0013] In one embodiment, the road point cloud data includes the position coordinates, radial velocity and echo intensity of each target point; and any point cloud feature includes at least one of the number of point clouds, the horizontal position coordinate of the point cloud centroid, the vertical position coordinate of the point cloud centroid, the ratio of the horizontal and vertical coordinates of the point cloud centroid, the relative radar distance of the point cloud centroid, the radial velocity of the point cloud centroid, the radial velocity variance of the point cloud centroid, the radial velocity variation difference of the point cloud centroid, the minimum value of the point cloud echo intensity, the maximum value of the point cloud echo intensity, the mean value of the point cloud echo intensity, the variance of the point cloud echo intensity, the horizontal position coordinate difference of the point cloud, the vertical position coordinate difference of the point cloud, the horizontal position coordinate variance of the point cloud, the vertical position coordinate variance of the point cloud and the point cloud density;
[0014] The step of extracting point cloud features from the road point cloud data to obtain point cloud features of the road point cloud data includes:
[0015] Determine the total number of target points in the road point cloud data as the point cloud number; and / or,
[0016] Determine the mean of the horizontal position coordinates of each target point in the road point cloud data as the horizontal position coordinates of the centroid of the point cloud; and / or,
[0017] Determine the mean value of the longitudinal position coordinates of each target point in the road point cloud data as the longitudinal position coordinates of the centroid of the point cloud; and / or,
[0018] Dividing the horizontal position coordinate of the point cloud mass center by the vertical position coordinate of the point cloud mass center to obtain the ratio of the horizontal and vertical coordinates of the point cloud mass center; and / or,
[0019] Obtaining the relative radar distance of the point cloud mass center according to the horizontal position coordinate of the point cloud mass center and the vertical position coordinate of the point cloud mass center; and / or,
[0020] Determine the average value of the radial velocity of each target point in the road point cloud data as the radial velocity of the point cloud centroid; and / or,
[0021] According to the radial velocity of each target point in the road point cloud data and the radial velocity of the point cloud centroid, the radial velocity variance of the point cloud centroid is obtained; and / or,
[0022] Determine the difference between the radial velocity of the target point with the largest value and the radial velocity of the target point with the smallest value in the road point cloud data as the radial velocity change difference of the point cloud mass center; and / or,
[0023] Determine the echo intensity of the target point with the smallest value in the road point cloud data as the minimum value of the point cloud echo intensity; and / or,
[0024] Determine the echo intensity of the target point with the largest value in the road point cloud data as the maximum value of the point cloud echo intensity; and / or,
[0025] Determine the mean value of the echo intensity of each target point in the road point cloud data as the point cloud echo intensity mean value;
[0026] Obtaining the point cloud echo intensity variance according to the echo intensity of each target point in the road point cloud data and the point cloud echo intensity mean;
[0027] Subtracting the horizontal position coordinate of the target point with the largest value from the horizontal position coordinate of the target point with the smallest value in the road point cloud data to obtain the horizontal position coordinate difference of the point cloud;
[0028] Subtracting the longitudinal position coordinate of the target point with the largest value from the longitudinal position coordinate of the target point with the smallest value in the road point cloud data to obtain the longitudinal position coordinate difference of the point cloud;
[0029] Obtaining the variance of the point cloud horizontal position coordinates according to the horizontal position coordinates of each target point in the road point cloud data and the horizontal position coordinates of the point cloud centroid;
[0030] Obtaining the variance of the vertical position coordinates of the point cloud according to the vertical position coordinates of each target point in the road point cloud data and the vertical position coordinates of the centroid of the point cloud;
[0031] The point cloud density is obtained according to the point cloud horizontal position coordinate difference, the point cloud vertical position coordinate difference and the number of point clouds.
[0032] In one embodiment, determining the mean of the horizontal position coordinates of each target point in the road point cloud data as the horizontal position coordinate of the centroid of the point cloud includes:
[0033] The horizontal position coordinates of the centroid of the point cloud are obtained by the following formula:
[0034]
[0035] in, is the horizontal position coordinate of the centroid of the point cloud, x i is the horizontal position coordinate of the target point i in the road point cloud data, i∈[1,N], N is the total number of target points in the road point cloud data;
[0036] The step of determining the mean value of the longitudinal position coordinates of each target point in the road point cloud data as the longitudinal position coordinate of the centroid of the point cloud includes:
[0037] The vertical position coordinate of the centroid of the point cloud is obtained by the following formula:
[0038]
[0039] in, is the vertical position coordinate of the centroid of the point cloud, y i is the vertical position coordinate of the target point i in the road point cloud data;
[0040] The step of dividing the horizontal position coordinate of the point cloud centroid by the vertical position coordinate of the point cloud centroid to obtain the ratio of the horizontal and vertical coordinates of the point cloud centroid includes:
[0041] The ratio of the horizontal and vertical coordinates of the point cloud centroid is obtained by the following formula:
[0042]
[0043] Wherein, σ is the ratio of the horizontal and vertical coordinates of the centroid of the point cloud;
[0044] The step of obtaining the relative radar distance of the point cloud mass center according to the horizontal position coordinate of the point cloud mass center and the vertical position coordinate of the point cloud mass center comprises:
[0045] The relative distance of the point cloud centroid to the radar is obtained by the following formula:
[0046]
[0047] Wherein, R is the distance of the point cloud centroid relative to the radar;
[0048] The step of determining the average value of the radial velocity of each target point in the road point cloud data as the radial velocity of the point cloud centroid includes:
[0049] The radial velocity of the point cloud centroid is obtained by:
[0050]
[0051] in, is the radial velocity of the point cloud centroid, v i is the radial velocity of the target point i in the road point cloud data;
[0052] The step of obtaining the radial velocity variance of the point cloud centroid according to the radial velocity of each target point in the road point cloud data and the radial velocity of the point cloud centroid comprises:
[0053] The radial velocity variance of the point cloud centroid is obtained by the following formula:
[0054]
[0055] Among them, δ v is the radial velocity variance of the point cloud centroid, v i is the radial velocity of the target point i in the road point cloud data, is the radial velocity of the point cloud centroid;
[0056] The step of determining the difference between the radial velocity of the target point with the largest value and the radial velocity of the target point with the smallest value in the road point cloud data as the radial velocity change difference of the point cloud mass center includes:
[0057] The radial velocity variation difference of the point cloud mass center is obtained by the following formula:
[0058] a v =v max -v min ;
[0059] Among them, a v is the radial velocity variation difference of the point cloud mass center, v max is the radial velocity of the target point with the largest value in the road point cloud data, v min The radial velocity of the target point with the smallest value in the road point cloud data;
[0060] The step of determining the mean value of the echo intensity of each target point in the road point cloud data as the point cloud echo intensity mean value comprises:
[0061] The mean value of the point cloud echo intensity is obtained by the following formula:
[0062]
[0063] in, is the mean value of the point cloud echo intensity, dB i is the echo intensity of target point i;
[0064] The step of obtaining the point cloud echo intensity variance according to the echo intensity of each target point in the road point cloud data and the point cloud echo intensity mean value includes:
[0065] The point cloud echo intensity variance is obtained by the following formula:
[0066]
[0067] Among them, δ dB is the point cloud echo intensity variance, dB i is the echo intensity of target point i in the road point cloud data, and dB is the mean value of the point cloud echo intensity;
[0068] Subtracting the horizontal position coordinate of the target point with the largest value from the horizontal position coordinate of the target point with the smallest value in the road point cloud data to obtain the horizontal position coordinate difference of the point cloud includes:
[0069] a x =x max -x min ;
[0070] Among them, a x is the horizontal position coordinate difference of the point cloud, x max is the horizontal position coordinate of the target point with the largest value in the road point cloud data, x min is the horizontal position coordinate of the target point with the largest value in the road point cloud data;
[0071] Subtracting the longitudinal position coordinate of the target point with the largest value from the longitudinal position coordinate of the target point with the smallest value in the road point cloud data to obtain the longitudinal position coordinate difference of the point cloud includes:
[0072] The vertical position coordinate difference of the point cloud is obtained by the following formula:
[0073] a y =y max -y min ;
[0074] Among them, a y is the vertical position coordinate difference of the point cloud, y max is the vertical position coordinate of the target point with the largest value in the road point cloud data, y min The vertical position coordinate of the target point with the largest value in the road point cloud data;
[0075] The step of obtaining the variance of the point cloud horizontal position coordinates according to the horizontal position coordinates of each target point in the road point cloud data and the horizontal position coordinates of the point cloud centroid comprises:
[0076] The horizontal position coordinate variance of the point cloud is obtained by the following formula:
[0077]
[0078] Among them, δ xis the horizontal position coordinate variance of the point cloud, x i is the horizontal position coordinate of the target point i in the road point cloud data, is the horizontal position coordinate of the centroid of the point cloud;
[0079] The method of obtaining the point cloud longitudinal position coordinate variance according to the longitudinal position coordinate of each target point in the road point cloud data and the longitudinal position coordinate of the point cloud centroid comprises:
[0080] The vertical position coordinate variance of the point cloud is obtained by the following formula:
[0081]
[0082] Among them, δ y is the vertical position coordinate variance of the point cloud, y i is the vertical position coordinate of the target point i in the road point cloud data, is the vertical position coordinate of the centroid of the point cloud;
[0083] The step of obtaining the point cloud density according to the horizontal position coordinate difference of the point cloud, the vertical position coordinate difference of the point cloud and the number of point clouds includes:
[0084] The point cloud density is obtained by the following formula:
[0085]
[0086] Where ρ is the point cloud density, α x is the horizontal position coordinate difference of the point cloud, α y is the vertical position coordinate difference of the point cloud.
[0087] In one embodiment, the HRRP of the road point cloud data includes the amplitude of each target point in the road point cloud data;
[0088] The HRRP geometric features include at least one of the equivalent scattering center number, equivalent target size, HRRP signal entropy, HRRP signal standard deviation, HRRP signal deviation, HRRP signal irregularity, HRRP signal echo energy, HRRP signal energy ratio, HRRP signal skewness and HRRP signal kurtosis;
[0089] The equivalent scattering center number is obtained by the following formula:
[0090]
[0091] Among them, Num scais the number of equivalent scattering centers, y(i) is the amplitude of target point i, m is the average amplitude of each target point in the road point cloud data, N is the total number of target points in the road point cloud data, when y(i)-m≥0, ε(y(i)-m)=1, when y(i)-m<0, ε(y(i)-m)=0;
[0092] The equivalent target size is obtained by the following formula:
[0093] Obtaining the equivalent target size according to the position coordinates of the target point with the largest amplitude and the position coordinates of the target point with the smallest amplitude in the HRRP;
[0094] The HRRP signal entropy is obtained by the following formula:
[0095]
[0096] Wherein, Entropy is the HRRP signal entropy;
[0097] The standard deviation of the HRRP signal is obtained by the following formula:
[0098]
[0099] Wherein, Std is the standard deviation of the HRRP signal;
[0100] The HRRP signal deviation is obtained by the following formula:
[0101]
[0102] Deviation is the HRRP signal deviation;
[0103] The HRRP signal irregularity is obtained by the following formula:
[0104]
[0105] Where Irr is the irregularity of the HRRP signal, y(i-1) is the amplitude of the target point i-1 before the target point i, and y(i+1) is the amplitude of the target point i+1 after the target point i;
[0106] The HRRP signal echo energy is obtained by the following formula:
[0107]
[0108] Wherein, TP is the HRRP signal echo energy;
[0109] The HRRP signal energy ratio is obtained by:
[0110] Sorting the amplitudes of the target points in the HRRP from large to small to obtain a sorted HRRP; and dividing the sum of the squares of the amplitudes of the first specified number of target points in the sorted HRRP by the sum of the squares of the amplitudes of the target points in the HRRP to obtain the HRRP signal energy ratio, wherein the specified number is equal to the average value of the amplitudes of the target points;
[0111] The HRRP signal skewness is obtained by the following formula:
[0112]
[0113] Wherein, Skewness is the skewness of the HRRP signal, and μ is the average value of the absolute value of the amplitude of each target point;
[0114] The HRRP signal kurtosis is obtained by the following formula:
[0115]
[0116] Wherein, urtosis is the kurtosis of the HRRP signal.
[0117] In one embodiment, the HRRP power spectrum characteristics of the road point cloud data are obtained by:
[0118] For the amplitude of any target point in the HRRP of the road point cloud data, Fourier transform is performed on the amplitude of the target point to obtain the frequency domain amplitude of the target point; and,
[0119] Determine the square of the frequency domain amplitude of the target point as the power spectrum of the target point;
[0120] The power spectrum of each target point in the HRRP of the road point cloud data is determined as the HRRP power spectrum feature of the road point cloud data.
[0121] A second aspect of the present disclosure provides an electronic device, comprising a processor and a memory, wherein the processor and the memory are connected via a bus;
[0122] The memory stores a computer program, and the processor is configured to perform the following operations based on the computer program:
[0123] At specified intervals, obtaining a specified number of frames of road point cloud data, wherein the road point cloud data is obtained by detecting the target road using a millimeter wave radar;
[0124] For any frame of road point cloud data, project the road point cloud data on the radial distance of the millimeter wave radar to obtain a high-resolution distance signal HRRP of the road point cloud data, and extract point cloud features from the road point cloud data to obtain point cloud features of each of the road point cloud data; and,
[0125] Obtaining a point cloud feature set of the road point cloud data according to the point cloud features, HRRP geometric features, and HRRP power spectrum features of the road point cloud data, wherein the HRRP geometric features and the HRRP power spectrum features are obtained by extracting features from the HRRP of the road point cloud data;
[0126] Obtaining a point cloud feature sequence according to the point cloud feature set of the road point cloud data of the specified number of frames;
[0127] The point cloud feature sequence is input into a pre-trained target classification model to obtain the category of each target object in the road point cloud data of the specified number of frames.
[0128] In one embodiment, the road point cloud data includes the position coordinates, radial velocity and echo intensity of each target point; and any point cloud feature includes at least one of the number of point clouds, the horizontal position coordinate of the point cloud centroid, the vertical position coordinate of the point cloud centroid, the ratio of the horizontal and vertical coordinates of the point cloud centroid, the relative radar distance of the point cloud centroid, the radial velocity of the point cloud centroid, the radial velocity variance of the point cloud centroid, the radial velocity variation difference of the point cloud centroid, the minimum value of the point cloud echo intensity, the maximum value of the point cloud echo intensity, the mean value of the point cloud echo intensity, the variance of the point cloud echo intensity, the horizontal position coordinate difference of the point cloud, the vertical position coordinate difference of the point cloud, the horizontal position coordinate variance of the point cloud, the vertical position coordinate variance of the point cloud and the point cloud density;
[0129] The processor performs the point cloud feature extraction on the road point cloud data to obtain the point cloud features of the road point cloud data, and is specifically configured as follows:
[0130] Determine the total number of target points in the road point cloud data as the point cloud number; and / or,
[0131] Determine the mean of the horizontal position coordinates of each target point in the road point cloud data as the horizontal position coordinates of the centroid of the point cloud; and / or,
[0132] Determine the mean value of the longitudinal position coordinates of each target point in the road point cloud data as the longitudinal position coordinates of the centroid of the point cloud; and / or,
[0133] Dividing the horizontal position coordinate of the point cloud mass center by the vertical position coordinate of the point cloud mass center to obtain the ratio of the horizontal and vertical coordinates of the point cloud mass center; and / or,
[0134] Obtaining the relative radar distance of the point cloud mass center according to the horizontal position coordinate of the point cloud mass center and the vertical position coordinate of the point cloud mass center; and / or,
[0135] Determine the average value of the radial velocity of each target point in the road point cloud data as the radial velocity of the point cloud centroid; and / or,
[0136] According to the radial velocity of each target point in the road point cloud data and the radial velocity of the point cloud centroid, the radial velocity variance of the point cloud centroid is obtained; and / or,
[0137] Determine the difference between the radial velocity of the target point with the largest value and the radial velocity of the target point with the smallest value in the road point cloud data as the radial velocity change difference of the point cloud mass center; and / or,
[0138] Determine the echo intensity of the target point with the smallest value in the road point cloud data as the minimum value of the point cloud echo intensity; and / or,
[0139] Determine the echo intensity of the target point with the largest value in the road point cloud data as the maximum value of the point cloud echo intensity; and / or,
[0140] Determine the mean value of the echo intensity of each target point in the road point cloud data as the point cloud echo intensity mean value;
[0141] Obtaining the point cloud echo intensity variance according to the echo intensity of each target point in the road point cloud data and the point cloud echo intensity mean;
[0142] Subtracting the horizontal position coordinate of the target point with the largest value from the horizontal position coordinate of the target point with the smallest value in the road point cloud data to obtain the horizontal position coordinate difference of the point cloud;
[0143] Subtracting the longitudinal position coordinate of the target point with the largest value from the longitudinal position coordinate of the target point with the smallest value in the road point cloud data to obtain the longitudinal position coordinate difference of the point cloud;
[0144] Obtaining the variance of the point cloud horizontal position coordinates according to the horizontal position coordinates of each target point in the road point cloud data and the horizontal position coordinates of the point cloud centroid;
[0145] Obtaining the variance of the vertical position coordinates of the point cloud according to the vertical position coordinates of each target point in the road point cloud data and the vertical position coordinates of the centroid of the point cloud;
[0146] The point cloud density is obtained according to the point cloud horizontal position coordinate difference, the point cloud vertical position coordinate difference and the number of point clouds.
[0147] In one embodiment, the processor determines the mean of the horizontal position coordinates of each target point in the road point cloud data as the horizontal position coordinates of the centroid of the point cloud, and is specifically configured as follows:
[0148] The horizontal position coordinates of the centroid of the point cloud are obtained by the following formula:
[0149]
[0150] in, is the horizontal position coordinate of the centroid of the point cloud, x i is the horizontal position coordinate of the target point i in the road point cloud data, i∈[1,N], N is the total number of target points in the road point cloud data;
[0151] The processor performs the step of determining the mean value of the longitudinal position coordinates of each target point in the road point cloud data as the longitudinal position coordinates of the centroid of the point cloud, and is specifically configured as follows:
[0152] The vertical position coordinate of the centroid of the point cloud is obtained by the following formula:
[0153]
[0154] in, is the vertical position coordinate of the centroid of the point cloud, y i is the vertical position coordinate of the target point i in the road point cloud data;
[0155] The processor performs the division of the horizontal position coordinate of the point cloud centroid by the vertical position coordinate of the point cloud centroid to obtain the ratio of the horizontal and vertical coordinates of the point cloud centroid, and is specifically configured as follows:
[0156] The ratio of the horizontal and vertical coordinates of the point cloud centroid is obtained by the following formula:
[0157]
[0158] Wherein, σ is the ratio of the horizontal and vertical coordinates of the centroid of the point cloud;
[0159] The processor executes the step of obtaining the relative radar distance of the point cloud mass center according to the horizontal position coordinate of the point cloud mass center and the vertical position coordinate of the point cloud mass center, and is specifically configured as follows:
[0160] The relative distance of the point cloud centroid to the radar is obtained by the following formula:
[0161]
[0162] Wherein, R is the distance of the point cloud centroid relative to the radar;
[0163] The processor performs the step of determining the average value of the radial velocity of each target point in the road point cloud data as the radial velocity of the point cloud centroid, and is specifically configured as follows:
[0164] The radial velocity of the point cloud centroid is obtained by:
[0165]
[0166] in, is the radial velocity of the point cloud centroid, v i is the radial velocity of the target point i in the road point cloud data;
[0167] The processor executes the step of obtaining the radial velocity variance of the point cloud centroid according to the radial velocity of each target point in the road point cloud data and the radial velocity of the point cloud centroid, and is specifically configured as follows:
[0168] The radial velocity variance of the point cloud centroid is obtained by the following formula:
[0169]
[0170] Among them, δ v is the radial velocity variance of the point cloud centroid, v i is the radial velocity of the target point i in the road point cloud data, is the radial velocity of the point cloud centroid;
[0171] The processor performs the step of determining the difference between the radial velocity of the target point with the largest value and the radial velocity of the target point with the smallest value in the road point cloud data as the radial velocity change difference of the point cloud mass center, and is specifically configured as follows:
[0172] The radial velocity variation difference of the point cloud mass center is obtained by the following formula:
[0173] a v =v max -v min ;
[0174] Among them, a v is the radial velocity variation difference of the point cloud mass center, v max is the radial velocity of the target point with the largest value in the road point cloud data, v min The radial velocity of the target point with the smallest value in the road point cloud data;
[0175] The processor performs the step of determining the mean value of the echo intensity of each target point in the road point cloud data as the point cloud echo intensity mean value, and is specifically configured as follows:
[0176] The mean value of the point cloud echo intensity is obtained by the following formula:
[0177]
[0178] in, is the mean value of the point cloud echo intensity, dB i is the echo intensity of target point i;
[0179] The processor executes the step of obtaining the point cloud echo intensity variance according to the echo intensity of each target point in the road point cloud data and the point cloud echo intensity mean, and is specifically configured as follows:
[0180] The point cloud echo intensity variance is obtained by the following formula:
[0181]
[0182] Among them, δ dB is the point cloud echo intensity variance, dB i is the echo intensity of target point i in the road point cloud data, is the mean value of the point cloud echo intensity;
[0183] The processor performs the step of subtracting the horizontal position coordinate of the target point with the largest value from the horizontal position coordinate of the target point with the smallest value in the road point cloud data to obtain the horizontal position coordinate difference of the point cloud, and is specifically configured as follows:
[0184] a x =x max -x min ;
[0185] Among them, a x is the horizontal position coordinate difference of the point cloud, x max is the horizontal position coordinate of the target point with the largest value in the road point cloud data, x min is the horizontal position coordinate of the target point with the largest value in the road point cloud data;
[0186] The processor performs the subtraction of the longitudinal position coordinate of the target point with the largest value in the road point cloud data from the longitudinal position coordinate of the target point with the smallest value to obtain the longitudinal position coordinate difference of the point cloud, and is specifically configured as follows:
[0187] The vertical position coordinate difference of the point cloud is obtained by the following formula:
[0188] a y =y max -y min ;
[0189] Among them, a y is the vertical position coordinate difference of the point cloud, y maxis the vertical position coordinate of the target point with the largest value in the road point cloud data, y min The vertical position coordinate of the target point with the largest value in the road point cloud data;
[0190] The processor executes the step of obtaining the variance of the point cloud horizontal position coordinates according to the horizontal position coordinates of each target point in the road point cloud data and the horizontal position coordinates of the point cloud centroid, and is specifically configured as follows:
[0191] The horizontal position coordinate variance of the point cloud is obtained by the following formula:
[0192]
[0193] Among them, δ x is the horizontal position coordinate variance of the point cloud, x i is the horizontal position coordinate of the target point i in the road point cloud data, is the horizontal position coordinate of the centroid of the point cloud;
[0194] The processor executes the step of obtaining the variance of the vertical position coordinates of the point cloud according to the vertical position coordinates of each target point in the road point cloud data and the vertical position coordinates of the centroid of the point cloud, and is specifically configured as follows:
[0195] The vertical position coordinate variance of the point cloud is obtained by the following formula:
[0196]
[0197] Among them, δ y is the vertical position coordinate variance of the point cloud, y i is the vertical position coordinate of the target point i in the road point cloud data, is the vertical position coordinate of the centroid of the point cloud;
[0198] The processor executes the step of obtaining the point cloud density according to the point cloud horizontal position coordinate difference, the point cloud vertical position coordinate difference and the number of point clouds, and is specifically configured as follows:
[0199] The point cloud density is obtained by the following formula:
[0200]
[0201] Where ρ is the point cloud density, α x is the horizontal position coordinate difference of the point cloud, α y is the vertical position coordinate difference of the point cloud.
[0202] In one embodiment, the HRRP of the road point cloud data includes the amplitude of each target point in the road point cloud data;
[0203] The HRRP geometric features include at least one of the equivalent scattering center number, equivalent target size, HRRP signal entropy, HRRP signal standard deviation, HRRP signal deviation, HRRP signal irregularity, HRRP signal echo energy, HRRP signal energy ratio, HRRP signal skewness and HRRP signal kurtosis;
[0204] The processor is further configured to:
[0205] The equivalent scattering center number is obtained by the following formula:
[0206]
[0207] Among them, Num sca is the number of equivalent scattering centers, y(i) is the amplitude of target point i, m is the average amplitude of each target point in the road point cloud data, N is the total number of target points in the road point cloud data, when y(i)-m≥0, ε(y(i)-m)=1, when y(i)-m<0, ε(y(i)-m)=0;
[0208] The equivalent target size is obtained by the following formula:
[0209] Obtaining the equivalent target size according to the position coordinates of the target point with the largest amplitude and the position coordinates of the target point with the smallest amplitude in the HRRP;
[0210] The HRRP signal entropy is obtained by the following formula:
[0211]
[0212] Wherein, Entropy is the HRRP signal entropy;
[0213] The standard deviation of the HRRP signal is obtained by the following formula:
[0214]
[0215] Wherein, Std is the standard deviation of the HRRP signal;
[0216] The HRRP signal deviation is obtained by the following formula:
[0217]
[0218] Deviation is the HRRP signal deviation;
[0219] The HRRP signal irregularity is obtained by the following formula:
[0220]
[0221] Where Irr is the irregularity of the HRRP signal, y(i-1) is the amplitude of the target point i-1 before the target point i, and y(i+1) is the amplitude of the target point i+1 after the target point i;
[0222] The HRRP signal echo energy is obtained by the following formula:
[0223]
[0224] Wherein, TP is the HRRP signal echo energy;
[0225] The HRRP signal energy ratio is obtained by:
[0226] Sorting the amplitudes of the target points in the HRRP from large to small to obtain a sorted HRRP; and dividing the sum of the squares of the amplitudes of the first specified number of target points in the sorted HRRP by the sum of the squares of the amplitudes of the target points in the HRRP to obtain the HRRP signal energy ratio, wherein the specified number is equal to the average value of the amplitudes of the target points;
[0227] The HRRP signal skewness is obtained by the following formula:
[0228]
[0229] Wherein, Skewness is the skewness of the HRRP signal, and μ is the average value of the absolute value of the amplitude of each target point;
[0230] The HRRP signal kurtosis is obtained by the following formula:
[0231]
[0232] Wherein, urtosis is the kurtosis of the HRRP signal.
[0233] In one embodiment, the processor is further configured to:
[0234] The HRRP power spectrum characteristics of the road point cloud data are obtained by:
[0235] For the amplitude of any target point in the HRRP of the road point cloud data, Fourier transform is performed on the amplitude of the target point to obtain the frequency domain amplitude of the target point; and,
[0236] Determine the square of the frequency domain amplitude of the target point as the power spectrum of the target point;
[0237] The power spectrum of each target point in the HRRP of the road point cloud data is determined as the HRRP power spectrum feature of the road point cloud data.
[0238] According to a third aspect provided by an embodiment of the present disclosure, a computer storage medium is provided, wherein the computer storage medium stores a computer program, and the computer program is used to execute the method as described in the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS
[0239] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0240] Figure 1 This is one of the schematic diagrams of applicable scenarios in one embodiment of the present disclosure;
[0241] Figure 2 This is a second schematic diagram of an applicable scenario in an embodiment of the present disclosure;
[0242] Figure 3 This is one of the flow charts of a method for multi-target recognition in a complex traffic environment according to an embodiment of the present disclosure;
[0243] Figure 4 A schematic diagram of road point cloud data according to an embodiment of the present disclosure;
[0244] Figure 5 A schematic diagram of a process of determining the road point cloud data according to an embodiment of the present disclosure;
[0245] Figure 6 is a schematic diagram of a road image according to an embodiment of the present disclosure;
[0246] Figure 7 A schematic diagram of a process for training a target classification model according to an embodiment of the present disclosure;
[0247] Figure 8 is a schematic diagram of a decision tree-based SVM model according to an embodiment of the present disclosure;
[0248] Fig. 9 A multi-target recognition device in a complex traffic environment according to an embodiment of the present disclosure;
[0249] Fig.10 The figure is a schematic diagram of the structure of an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION
[0250] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present disclosure.
[0251] In the embodiments of the present disclosure, the term "and / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent three situations: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects before and after are in an "or" relationship.
[0252] The application scenarios described in the embodiments of the present disclosure are intended to more clearly illustrate the technical solutions of the embodiments of the present disclosure, and do not constitute a limitation on the technical solutions provided by the embodiments of the present disclosure. It is known to those skilled in the art that with the emergence of new application scenarios, the technical solutions provided by the embodiments of the present disclosure are also applicable to similar technical problems. In the description of the present disclosure, unless otherwise specified, the meaning of "multiple" is two or more.
[0253] In the prior art, the method of realizing traffic road vehicle recognition is to obtain the classification result based on the point cloud data generated by a single detection of the target. However, in the actual road target classification task, the classification effect of this method is easily affected by abnormal detection results, and the accuracy of target recognition based on only a single frame of point cloud data is low in a crowded and complex traffic environment.
[0254] Therefore, the present disclosure provides a multi-target recognition method in a complex traffic environment, which first extracts the point cloud features, HRRP geometric features and HRRP power spectrum features of single-frame road point cloud data to determine the point cloud feature set of single-frame road point cloud data, and then combines the point cloud feature sets of multi-frame road point cloud data to obtain a point cloud feature sequence. And target recognition is performed based on the point cloud feature sequence to obtain the category of each target object. Therefore, in the embodiment of the application, multiple features in a single-frame road point cloud data are extracted to accurately obtain the point cloud feature set of each target object; and a high-dimensional multi-frame point cloud feature sequence is obtained by fusing the point cloud feature set of multi-frame road point cloud data on a time series, so that more detailed information of each target object in a complex traffic environment can be captured to improve the accuracy of multi-target recognition in a complex traffic environment. Below, the scheme of the present disclosure is introduced in detail in conjunction with the accompanying drawings.
[0255] like Figure 1As shown, a schematic diagram of an application scenario of a multi-target recognition method in a complex traffic environment includes a millimeter wave radar 110, a server 120 and a terminal device 130. Figure 1 As shown, the application scenario is described by taking the electronic device as a server as an example. However, the embodiment of the present application does not limit the electronic device, and the electronic device can be a server and a terminal device, etc., which can be set according to actual conditions. The server 120 can be implemented by a single server or by multiple servers. The server 120 can be implemented by a physical server or by a virtual server.
[0256] In a possible application scenario, the millimeter wave radar 110 performs real-time detection on the target road to obtain the road point cloud data of the target road. The server 120 obtains the road point cloud data of a specified number of frames at a specified time interval. Then, for any frame of road point cloud data, the server 120 projects the road point cloud data on the radial distance of the millimeter wave radar to obtain the high-resolution distance signal HRRP of the road point cloud data, and extracts point cloud features from the road point cloud data to obtain the point cloud features of the road point cloud data. And based on the point cloud features of the road point cloud data and the HRRP geometric features, the server 120 extracts the point cloud features of the road point cloud data. The server 120 obtains a point cloud feature sequence according to the point cloud feature set of the road point cloud data of the specified frame number; the point cloud feature sequence is input into a pre-trained target classification model to obtain the category of each target object in the road point cloud data of the specified frame number, and the category of each target object in the road point cloud data of the specified frame number is sent to the terminal device 130 for display.
[0257] like Figure 2As shown, it is a schematic diagram of another application scenario in the embodiment of the present application, and the application scenario includes a millimeter wave radar 110 and a terminal device 130, wherein the millimeter wave radar 110 performs real-time detection on the target road to obtain the road point cloud data of the target road, and the terminal device 130 obtains a specified number of frames of road point cloud data at specified intervals, and then the terminal device 130 projects the road point cloud data on the radial distance of the millimeter wave radar for any frame of road point cloud data to obtain a high-resolution distance signal HRRP of the road point cloud data, and extracts point cloud features from the road point cloud data to obtain point cloud features of the road point cloud data; and according to The point cloud features, HRRP geometric features and HRRP power spectrum features of the road point cloud data are used to obtain a point cloud feature set of the road point cloud data, wherein the HRRP geometric features and the HRRP power spectrum features are obtained by extracting features from the HRRP of the road point cloud data; the terminal device 130 obtains a point cloud feature sequence according to the point cloud feature set of the road point cloud data of the specified frame number; the point cloud feature sequence is input into a pre-trained target classification model to obtain the category of each target object in the road point cloud data of the specified frame number, and the category of each target object in the road point cloud data of the specified frame number is displayed.
[0258] in, Figure 1 The server 120 and the terminal device 130 can exchange information through a communication network, wherein the communication mode adopted by the communication network can be divided into a wireless communication mode or a wired communication mode.
[0259] Exemplarily, the server 120 can access the network through cellular mobile communication technology to communicate with the terminal device 130, wherein the cellular mobile communication technology, for example, includes the fifth generation mobile communication (5th Generation Mobile Networks, 5G) technology.
[0260] Optionally, the server 120 may access the network and communicate with the terminal device 130 via short-range wireless communication, wherein the short-range wireless communication method, for example, includes Wireless Fidelity (Wi-Fi) technology.
[0261] Among them, the description in this application only details a single millimeter wave radar 110, a single server 120, and a single terminal device 130, but those skilled in the art should understand that the millimeter wave radar 110, server 120, and terminal device 130 shown are intended to represent the operations of the millimeter wave radar 110, server 120, and terminal device 130 involved in the technical solution of this application. It does not imply that there is a limit on the number, type, or location of the millimeter wave radar 110, server 120, and terminal device 130. It should be noted that if additional modules are added to the illustrated environment or individual modules are removed therefrom, the underlying concept of the exemplary embodiment of this application will not be changed.
[0262] It should be noted that the multi-target recognition method in complex traffic environments proposed in this application is not only applicable to Figure 1 and Figure 2 The application scenario shown is also applicable to any multi-target recognition device in a complex traffic environment.
[0263] The following describes the multi-target recognition method in a complex traffic environment in combination with the application scenarios described above and with reference to the accompanying drawings. It should be noted that the above application scenarios are only shown to facilitate understanding of the methods and principles of the present application, and the implementation methods of the present application are not limited in this regard.
[0264] like Figure 3 FIG. 1 is a flow chart of a method for multi-target recognition in a complex traffic environment disclosed in the present invention, which may include the following steps:
[0265] Step 301: acquiring road point cloud data of a specified number of frames at specified time intervals, wherein the road point cloud data is obtained by detecting the target road using a millimeter wave radar;
[0266] like Figure 4 As shown, it is a schematic diagram of road point cloud data. Since there are only vehicles in the target road, each target point in the road point cloud data is a target point of the vehicle.
[0267] Among them, the road point cloud data in the embodiment of the present application is obtained by performing a two-dimensional FFT (Fast Fourier transform) conversion on the original signal sampling data containing the activity information of the target object obtained by the millimeter wave radar to obtain two-dimensional spectrum data containing the relative speed and relative distance information of the target object; then the two-dimensional spectrum data is peak detected by the CFAR (Constant False-Alarm Rate) algorithm to obtain peak data containing the speed and distance of the target object; then the extracted peak data is subjected to DOA (Direction of arrival) angle acquisition to obtain point cloud data containing relative distance, direction angle and pitch angle; and the spatial coordinate information of the point cloud data is obtained by trigonometric function theorem, and the motion feature set is obtained by calculation to obtain the road point cloud data.
[0268] It should be noted that the method of obtaining road point cloud data from the original signal obtained by the millimeter wave radar in the embodiment of the present application can be set according to the actual situation. The method described above is only used as an example and does not limit the method of obtaining road point cloud data. In addition, the target road, the specified time length and the specified number of frames in the embodiment of the present application can be set according to the actual situation. The embodiment of the present application does not limit the specified time length and the specified number of frames.
[0269] Step 302: for any frame of road point cloud data, project the road point cloud data on the radial distance of the millimeter wave radar to obtain a high-resolution distance signal HRRP of the road point cloud data, and extract point cloud features from the road point cloud data to obtain point cloud features of the road point cloud data;
[0270] In the embodiment of the present application, the method of projecting to obtain the road point cloud data is as follows: if the road point cloud data of the bth frame is a set P = {p1, ..., p n}, where P i =(d i ,m i ), d i Represents the spatial position coordinates of target point i, m i Represents the state vector of target point i, which includes the radar cross-sectional area, signal-to-noise ratio, etc. of target point i. The spatial position coordinates of the road point cloud data are converted into a three-dimensional map of RDA (Radar Data Acquisition, radar data acquisition subsystem), and then the inverse Fourier transform is performed on it in the angle dimension and the speed dimension to obtain the HRRP of the road point cloud data of the bth frame.
[0271] The projection method described above in the embodiment of the present application is only used as an example for illustration and does not limit the method for obtaining HRRP. The specific method can be set according to actual conditions.
[0272] The road point cloud data includes the position coordinates, radial velocity and echo intensity of each target point; and any point cloud feature includes at least one of the number of point clouds, the horizontal position coordinate of the point cloud centroid, the vertical position coordinate of the point cloud centroid, the ratio of the horizontal and vertical coordinates of the point cloud centroid, the relative radar distance of the point cloud centroid, the radial velocity of the point cloud centroid, the radial velocity variance of the point cloud centroid, the radial velocity variation difference of the point cloud centroid, the minimum value of the point cloud echo intensity, the maximum value of the point cloud echo intensity, the mean value of the point cloud echo intensity, the variance of the point cloud echo intensity, the horizontal position coordinate difference of the point cloud, the vertical position coordinate difference of the point cloud, the horizontal position coordinate variance of the point cloud, the vertical position coordinate variance of the point cloud and the point cloud density; below, the determination method of each point cloud feature is described respectively:
[0273] 1. Point cloud quantity: The total number of target points in the road point cloud data is determined as the point cloud quantity.
[0274] 2. Horizontal position coordinates of the point cloud centroid: The mean of the horizontal position coordinates of each target point in the road point cloud data is determined as the horizontal position coordinates of the point cloud centroid. The horizontal position coordinates of the point cloud centroid can be obtained by formula (1):
[0275]
[0276] in, is the horizontal position coordinate of the centroid of the point cloud, x i is the horizontal position coordinate of the target point i in the road point cloud data, i∈[1,N], and N is the total number of target points in the road point cloud data.
[0277] 3. The vertical position coordinate of the centroid of the point cloud: The average of the vertical position coordinates of each target point in the road point cloud data is determined as the vertical position coordinate of the centroid of the point cloud. The vertical position coordinate of the centroid of the point cloud can be obtained by formula (2):
[0278]
[0279] in, is the vertical position coordinate of the centroid of the point cloud, y i is the vertical position coordinate of the target point i in the road point cloud data.
[0280] 4. The ratio of the horizontal and vertical coordinates of the point cloud mass center: Divide the horizontal position coordinate of the point cloud mass center by the vertical position coordinate of the point cloud mass center to obtain the ratio of the horizontal and vertical coordinates of the point cloud mass center. The ratio of the horizontal and vertical coordinates of the point cloud mass center can be obtained by formula (3):
[0281]
[0282] Wherein, σ is the ratio of the horizontal and vertical coordinates of the centroid of the point cloud.
[0283] 5. The relative radar distance of the point cloud mass center: According to the horizontal position coordinate of the point cloud mass center and the vertical position coordinate of the point cloud mass center, the relative radar distance of the point cloud mass center is obtained. The relative radar distance of the point cloud mass center can be obtained by formula (4):
[0284]
[0285] Wherein, R is the distance of the point cloud centroid relative to the radar.
[0286] 6. Point cloud centroid radial velocity: The average value of the radial velocity of each target point in the road point cloud data is determined as the point cloud centroid radial velocity. The point cloud centroid radial velocity can be obtained by formula (5):
[0287]
[0288] in, is the radial velocity of the point cloud centroid, v i is the radial velocity of target point i in the road point cloud data.
[0289] 7. Point cloud centroid radial velocity variance: According to the radial velocity of each target point in the road point cloud data and the radial velocity of the point cloud centroid, the point cloud centroid radial velocity variance is obtained. The point cloud centroid radial velocity variance can be obtained by formula (6):
[0290]
[0291] Among them, δ v is the radial velocity variance of the point cloud centroid, v i is the radial velocity of the target point i in the road point cloud data, is the radial velocity of the point cloud centroid.
[0292] 8. Point cloud mass center radial velocity change difference: The difference between the radial velocity of the target point with the largest value and the radial velocity of the target point with the smallest value in the road point cloud data is determined as the point cloud mass center radial velocity change difference. The point cloud mass center radial velocity change difference can be obtained by formula (7):
[0293] a v =v max -v min ……(7);
[0294] Among them, a vis the radial velocity variation difference of the point cloud mass center, v max is the radial velocity of the target point with the largest value in the road point cloud data, v min is the radial velocity of the target point with the smallest value in the road point cloud data.
[0295] 9. Minimum value of point cloud echo intensity: The echo intensity of the target point with the smallest value in the road point cloud data is determined as the minimum value of the point cloud echo intensity. The minimum value of the point cloud echo intensity can be obtained by formula (8):
[0296] dB min =min(dB i )……(8);
[0297] Where, dB min is the minimum value of the point cloud echo intensity, dB i is the echo intensity of target point i in the road point cloud data, i∈[1,N].
[0298] 10. Maximum value of point cloud echo intensity: The echo intensity of the target point with the largest value in the road point cloud data is determined as the maximum value of the point cloud echo intensity. The maximum value of the point cloud echo intensity can be obtained by formula (9):
[0299] dB max =max(dB i )……(9);
[0300] Where, dB max is the maximum value of the point cloud echo intensity.
[0301] 11. Point cloud echo intensity mean: The mean of the echo intensity of each target point in the road point cloud data is determined as the point cloud echo intensity mean. The point cloud echo intensity mean can be obtained by formula (10):
[0302]
[0303] in, is the mean value of the point cloud echo intensity, dB i is the echo intensity of target point i.
[0304] 12. Point cloud echo intensity variance: The point cloud echo intensity variance is obtained according to the echo intensity of each target point in the road point cloud data and the mean value of the point cloud echo intensity. The point cloud echo intensity variance can be obtained by formula (11):
[0305]
[0306] Among them, δ dBis the point cloud echo intensity variance, dB i is the echo intensity of target point i in the road point cloud data, is the mean value of the point cloud echo intensity 13. Point cloud horizontal position coordinate difference: subtract the horizontal position coordinate of the target point with the largest value from the horizontal position coordinate of the target point with the smallest value in the road point cloud data to obtain the point cloud horizontal position coordinate difference.
[0307] 13. Point cloud horizontal position coordinate difference: Subtract the horizontal position coordinate of the target point with the largest value from the horizontal position coordinate of the target point with the smallest value in the road point cloud data to obtain the point cloud horizontal position coordinate difference. The point cloud horizontal position coordinate difference can be obtained by formula (12):
[0308] a x =x max -x min ……(12);
[0309] Among them, a x is the horizontal position coordinate difference of the point cloud, x max is the horizontal position coordinate of the target point with the largest value in the road point cloud data, x min is the horizontal position coordinate of the target point with the largest value in the road point cloud data.
[0310] 14. Point cloud longitudinal position coordinate difference: Subtract the longitudinal position coordinate of the target point with the largest value from the longitudinal position coordinate of the target point with the smallest value in the road point cloud data to obtain the point cloud longitudinal position coordinate difference. The point cloud longitudinal position coordinate difference can be obtained by formula (13):
[0311] a y =y max -y min ……(13);
[0312] Among them, a y is the vertical position coordinate difference of the point cloud, y max is the vertical position coordinate of the target point with the largest value in the road point cloud data, y min It is the vertical position coordinate of the target point with the largest value in the road point cloud data.
[0313] 15. Point cloud horizontal position coordinate variance: According to the horizontal position coordinates of each target point in the road point cloud data and the horizontal position coordinates of the point cloud centroid, the point cloud horizontal position coordinate variance is obtained. The point cloud horizontal position coordinate variance can be obtained by formula (14):
[0314]
[0315] Among them, δ x is the horizontal position coordinate variance of the point cloud, xi is the horizontal position coordinate of the target point i in the road point cloud data, is the horizontal position coordinate of the centroid of the point cloud.
[0316] 16. Point cloud longitudinal position coordinate variance: According to the longitudinal position coordinates of each target point in the road point cloud data and the longitudinal position coordinates of the point cloud centroid, the point cloud longitudinal position coordinate variance is obtained. The point cloud longitudinal position coordinate variance can be obtained by formula (15):
[0317]
[0318] Among them, δ y is the vertical position coordinate variance of the point cloud, y i is the vertical position coordinate of the target point i in the road point cloud data, is the vertical position coordinate of the centroid of the point cloud.
[0319] 17. Point cloud density: The point cloud density is obtained according to the horizontal position coordinate difference of the point cloud, the vertical position coordinate difference of the point cloud and the number of point clouds. The point cloud density can be obtained by formula (16):
[0320]
[0321] Where ρ is the point cloud density, α x is the horizontal position coordinate difference of the point cloud, α y is the vertical position coordinate difference of the point cloud.
[0322] Step 303: obtaining a point cloud feature set of each of the road point cloud data according to the point cloud features, HRRP geometric features, and HRRP power spectrum features of the road point cloud data, wherein the HRRP geometric features and the HRRP power spectrum features are obtained by extracting features from the HRRP of the road point cloud data;
[0323] The HRRP of the road point cloud data includes the amplitude of each target point in the road point cloud data; the HRRP geometric features include at least one of the number of equivalent scattering centers, equivalent target size, HRRP signal entropy, HRRP signal standard deviation, HRRP signal deviation, HRRP signal irregularity, HRRP signal echo energy, HRRP signal energy ratio, HRRP signal skewness and HRRP signal kurtosis;
[0324] Next, the method of extracting HRRP geometric features in the embodiment of the present application is first introduced:
[0325] 1. Number of equivalent scattering centers: The number of equivalent scattering centers can be obtained by formula (17):
[0326]
[0327] Among them, Num sca is the number of equivalent scattering centers, y(i) is the amplitude of target point i, m is the average amplitude of each target point in the road point cloud data, N is the total number of target points in the road point cloud data, when y(i)-m≥0, ε(y(i)-m)=1, when y(i)-m<0, ε(y(i)-m)=0.
[0328] 2. Equivalent target size: The equivalent target size is obtained according to the position coordinates of the target point with the largest amplitude and the position coordinates of the target point with the smallest amplitude in the HRRP.
[0329] In one embodiment, the horizontal position coordinate of the target point with the largest amplitude in the HRRP is subtracted from the horizontal position coordinate of the target point with the smallest amplitude to obtain the length of the equivalent target; and the vertical position coordinate of the target point with the largest amplitude in the HRRP is subtracted from the vertical position coordinate of the target point with the smallest amplitude to obtain the width of the equivalent target, and the length of the equivalent target and the width of the equivalent target are determined as the equivalent target size.
[0330] 3. HRRP signal entropy: The HRRP signal entropy is obtained by formula (18):
[0331]
[0332] Wherein, Entropy is the HRRP signal entropy.
[0333] 4. HRRP signal standard deviation: The HRRP signal standard deviation is obtained by formula (19):
[0334]
[0335] Wherein, Std is the standard deviation of the HRRP signal.
[0336] 5. HRRP signal deviation: The HRRP signal deviation is obtained by formula (20):
[0337]
[0338] Deviation is the HRRP signal deviation.
[0339] 6. HRRP signal irregularity: The HRRP signal irregularity is obtained by formula (21):
[0340]
[0341] Where Irr is the irregularity of the HRRP signal, y(i-1) is the amplitude of the target point i-1 before the target point i, and y(i+1) is the amplitude of the target point i+1 after the target point i.
[0342] 7. HRRP signal echo energy: The HRRP signal echo energy is obtained by formula (22):
[0343]
[0344] Wherein, TP is the HRRP signal echo energy.
[0345] 8. HRRP signal energy ratio: sort the amplitudes of the target points in the HRRP from large to small to obtain the sorted HRRP; and divide the sum of the squares of the amplitudes of the first specified number of target points in the sorted HRRP by the sum of the squares of the amplitudes of the target points in the HRRP to obtain the HRRP signal energy ratio, wherein the specified number is equal to the average value of the amplitudes of the target points. The HRRP signal energy ratio is obtained by formula (23):
[0346]
[0347] Among them, P m is the HRRP signal energy ratio, and x(i) is the amplitude of the target point i in the sorted HRRP.
[0348] 9. HRRP signal skewness: The HRRP signal skewness is obtained by formula (24):
[0349]
[0350] Wherein, Skewness is the skewness of the HRRP signal, and μ is the average value of the absolute value of the amplitude of each target point. The average value of the absolute value of the amplitude of each target point can be obtained by formula (25):
[0351]
[0352] Wherein, μ is the average value of the absolute values of the amplitudes of the target points.
[0353] 10. HRRP signal kurtosis: The HRRP signal kurtosis is obtained by formula (26):
[0354]
[0355] Wherein, urtosis is the kurtosis of the HRRP signal.
[0356] Next, the method of determining the HRRP power spectrum characteristics of the road point cloud data is introduced, such as Figure 5 As shown, it is a schematic diagram of the process of determining the road point cloud data, which includes the following steps:
[0357] Step 501: for the amplitude of any target point in the HRRP of the road point cloud data, perform Fourier transform on the amplitude of the target point to obtain the frequency domain amplitude of the target point;
[0358] Step 502: Determine the square of the frequency domain amplitude of the target point as the power spectrum of the target point; wherein the power spectrum of the target point can be obtained by formula (27):
[0359] D(i)=w(i) 2 ……(27);
[0360] Where D(i) is the power spectrum of target point i, and w(i) is the frequency domain amplitude of target point i.
[0361] Step 503: Determine the power spectrum of each target point in the HRRP of the road point cloud data as the HRRP power spectrum feature of the road point cloud data.
[0362] Step 304: obtaining a point cloud feature sequence according to the point cloud feature set of the road point cloud data of the specified number of frames;
[0363] In one embodiment, the point cloud feature sequence can be obtained by the following two methods:
[0364] Method 1: The point cloud feature set of the road point cloud data of the specified number of frames is combined into a feature sequence, and the feature sequence is determined as the point cloud feature sequence.
[0365] Method 2: The point cloud feature set of the road point cloud data of the specified number of frames is formed into a feature sequence. For any feature in the feature sequence, the number of the features in the feature sequence is counted, and the number of the features in the feature sequence is divided by the total number of features in the feature sequence to obtain the probability density of the feature. The features in the feature sequence whose probability density is less than a preset threshold are deleted from the feature sequence to obtain the point cloud feature sequence.
[0366] Step 305: input the point cloud feature sequence into a pre-trained target classification model to obtain the category of each target object in the road point cloud data of the specified number of frames.
[0367] The target object in the embodiment of the present application may be a motor vehicle, a pedestrian, a non-motor vehicle, etc., and the specific setting may be made according to the actual situation, and the embodiment of the present application is not limited here.
[0368] like Figure 6 As shown in FIG. 1 , a road image corresponding to any frame of road point cloud data is shown. From the figure, it can be seen that the position of each target object in the road image and the category corresponding to each target object are shown.
[0369] The following is a brief introduction to the training method of the target classification model in the embodiment of the present application. Figure 7 As shown in FIG. 1 , a schematic diagram of the training process of the target classification model is shown, which may include the following steps:
[0370] Step 701: obtaining a training sample, wherein the training sample includes a point cloud feature sequence corresponding to a plurality of frames of road point cloud data and a standard position and annotated category of each target object in the plurality of frames of road point cloud data, and any road image includes an annotated position of a target object and an annotated category of the target object;
[0371] Step 702: input the training sample into the target classification model for detection, and obtain the predicted position and predicted category of each target object in the multi-frame road point cloud data in the training sample;
[0372] Step 703: Obtaining a first loss value based on the predicted position and the marked position of each target object in the multi-frame road point cloud data;
[0373] Step 704: obtaining a second loss value based on the predicted category and the labeled category of each target object in the multi-frame road point cloud data;
[0374] Step 705: Add the first loss value and the second loss value to obtain a total loss value;
[0375] Step 706: Determine whether the total loss value is less than the specified loss value, if so, execute step 707, if not, execute step 708;
[0376] Step 707: After adjusting the model parameters of the decision tree-based SVM model, return to step 702;
[0377] Step 708: End the training of the target classification model to obtain the trained target classification model.
[0378] The target classification model in the embodiment of the present application uses a SVM (Support Vector Machine) model based on a decision tree. The decision tree in the embodiment of the present application adopts an unbalanced binary tree architecture, such as Figure 8 As shown, the root node SVM1 divides the three types of targets into two categories: motor vehicles and pedestrians and non-motor vehicles; the second-level child node SVM2 further distinguishes pedestrians and non-motor vehicles.
[0379] It should be noted that the target classification model described above is a decision tree-based SVM model, which is only used for illustration and is not limited to the target classification model in the embodiments of the present application. The target classification model in the embodiments of the present application can be set according to actual conditions.
[0380] Based on the same disclosure concept, the multi-target recognition method in a complex traffic environment described above can also be implemented by a multi-target recognition device in a complex traffic environment. The effect of the multi-target recognition device in a complex traffic environment is similar to that of the aforementioned method, and will not be repeated here.
[0381] Fig. 9 It is a schematic diagram of the structure of a multi-target recognition device in a complex traffic environment according to an embodiment of the present disclosure.
[0382] like Fig. 9 As shown, the multi-target recognition device 900 in a complex traffic environment of the present invention may include a road point cloud data acquisition module 910, a point cloud feature extraction module 920, a point cloud feature set determination module 930, a point cloud feature sequence determination module 940 and a target object category determination module 950.
[0383] A road point cloud data acquisition module 910 is used to acquire a specified number of frames of road point cloud data at specified intervals, wherein the road point cloud data is obtained by detecting the target road using a millimeter wave radar;
[0384] The point cloud feature extraction module 920 is used to project the road point cloud data on the radial distance of the millimeter wave radar for any frame of road point cloud data to obtain a high-resolution distance signal HRRP of the road point cloud data, and to extract point cloud features from the road point cloud data to obtain point cloud features of the road point cloud data; and
[0385] A point cloud feature set determination module 930 is used to obtain a point cloud feature set of each of the road point cloud data according to the point cloud features, HRRP geometric features and HRRP power spectrum features of the road point cloud data, wherein the HRRP geometric features and the HRRP power spectrum features are obtained by extracting features from the HRRP of the road point cloud data;
[0386] A point cloud feature sequence determination module 940 is used to obtain a point cloud feature sequence according to a point cloud feature set of the road point cloud data of the specified number of frames;
[0387] The target object category determination module 950 is used to input the point cloud feature sequence into a pre-trained target classification model to obtain the category of each target object in the road point cloud data of the specified number of frames.
[0388] In one embodiment, the road point cloud data includes the position coordinates, radial velocity and echo intensity of each target point; and any point cloud feature includes at least one of the number of point clouds, the horizontal position coordinate of the point cloud centroid, the vertical position coordinate of the point cloud centroid, the ratio of the horizontal and vertical coordinates of the point cloud centroid, the relative radar distance of the point cloud centroid, the radial velocity of the point cloud centroid, the radial velocity variance of the point cloud centroid, the radial velocity variation difference of the point cloud centroid, the minimum value of the point cloud echo intensity, the maximum value of the point cloud echo intensity, the mean value of the point cloud echo intensity, the variance of the point cloud echo intensity, the horizontal position coordinate difference of the point cloud, the vertical position coordinate difference of the point cloud, the horizontal position coordinate variance of the point cloud, the vertical position coordinate variance of the point cloud and the point cloud density;
[0389] The point cloud feature extraction module 920 is specifically used for:
[0390] Determine the total number of target points in the road point cloud data as the point cloud number; and / or,
[0391] Determine the mean of the horizontal position coordinates of each target point in the road point cloud data as the horizontal position coordinates of the centroid of the point cloud; and / or,
[0392] Determine the mean value of the longitudinal position coordinates of each target point in the road point cloud data as the longitudinal position coordinates of the centroid of the point cloud; and / or,
[0393] Dividing the horizontal position coordinate of the point cloud mass center by the vertical position coordinate of the point cloud mass center to obtain the ratio of the horizontal and vertical coordinates of the point cloud mass center; and / or,
[0394] Obtaining the relative radar distance of the point cloud mass center according to the horizontal position coordinate of the point cloud mass center and the vertical position coordinate of the point cloud mass center; and / or,
[0395] Determine the average value of the radial velocity of each target point in the road point cloud data as the radial velocity of the point cloud centroid; and / or,
[0396] According to the radial velocity of each target point in the road point cloud data and the radial velocity of the point cloud centroid, the radial velocity variance of the point cloud centroid is obtained; and / or,
[0397] Determine the difference between the radial velocity of the target point with the largest value and the radial velocity of the target point with the smallest value in the road point cloud data as the radial velocity change difference of the point cloud mass center; and / or,
[0398] Determine the echo intensity of the target point with the smallest value in the road point cloud data as the minimum value of the point cloud echo intensity; and / or,
[0399] Determine the echo intensity of the target point with the largest value in the road point cloud data as the maximum value of the point cloud echo intensity; and / or,
[0400] Determine the mean value of the echo intensity of each target point in the road point cloud data as the point cloud echo intensity mean value;
[0401] Obtaining the point cloud echo intensity variance according to the echo intensity of each target point in the road point cloud data and the point cloud echo intensity mean;
[0402] Subtracting the horizontal position coordinate of the target point with the largest value from the horizontal position coordinate of the target point with the smallest value in the road point cloud data to obtain the horizontal position coordinate difference of the point cloud;
[0403] Subtracting the longitudinal position coordinate of the target point with the largest value from the longitudinal position coordinate of the target point with the smallest value in the road point cloud data to obtain the longitudinal position coordinate difference of the point cloud;
[0404] Obtaining the variance of the point cloud horizontal position coordinates according to the horizontal position coordinates of each target point in the road point cloud data and the horizontal position coordinates of the point cloud centroid;
[0405] Obtaining the variance of the vertical position coordinates of the point cloud according to the vertical position coordinates of each target point in the road point cloud data and the vertical position coordinates of the centroid of the point cloud;
[0406] The point cloud density is obtained according to the point cloud horizontal position coordinate difference, the point cloud vertical position coordinate difference and the number of point clouds.
[0407] In one embodiment, the point cloud feature extraction module 920 performs the step of determining the mean of the horizontal position coordinates of each target point in the road point cloud data as the horizontal position coordinates of the centroid of the point cloud, specifically for:
[0408] The horizontal position coordinates of the centroid of the point cloud are obtained by the following formula:
[0409]
[0410] in, is the horizontal position coordinate of the centroid of the point cloud, x i is the horizontal position coordinate of the target point i in the road point cloud data, i∈[1,N], N is the total number of target points in the road point cloud data;
[0411] The point cloud feature extraction module 920 performs the step of determining the mean value of the longitudinal position coordinates of each target point in the road point cloud data as the longitudinal position coordinates of the centroid of the point cloud, specifically for:
[0412] The vertical position coordinate of the centroid of the point cloud is obtained by the following formula:
[0413]
[0414] in, is the vertical position coordinate of the centroid of the point cloud, y i is the vertical position coordinate of the target point i in the road point cloud data;
[0415] The point cloud feature extraction module 920 performs the division of the horizontal position coordinate of the point cloud centroid by the vertical position coordinate of the point cloud centroid to obtain the ratio of the horizontal and vertical coordinates of the point cloud centroid, which is specifically used for:
[0416] The ratio of the horizontal and vertical coordinates of the point cloud centroid is obtained by the following formula:
[0417]
[0418] Wherein, σ is the ratio of the horizontal and vertical coordinates of the centroid of the point cloud;
[0419] The point cloud feature extraction module 920 performs the step of obtaining the relative radar distance of the point cloud centroid according to the horizontal position coordinate of the point cloud centroid and the vertical position coordinate of the point cloud centroid, and is specifically used for:
[0420] The relative distance of the point cloud centroid to the radar is obtained by the following formula:
[0421]
[0422] Wherein, R is the distance of the point cloud centroid relative to the radar;
[0423] The point cloud feature extraction module 920 performs the step of determining the average radial velocity of each target point in the road point cloud data as the point cloud centroid radial velocity, specifically for:
[0424] The radial velocity of the point cloud centroid is obtained by:
[0425]
[0426] in, is the radial velocity of the point cloud centroid, v i is the radial velocity of the target point i in the road point cloud data;
[0427] The point cloud feature extraction module 920 performs the step of obtaining the radial velocity variance of the point cloud centroid according to the radial velocity of each target point in the road point cloud data and the radial velocity of the point cloud centroid, which is specifically used for:
[0428] The radial velocity variance of the point cloud centroid is obtained by the following formula:
[0429]
[0430] Among them, δ v is the radial velocity variance of the point cloud centroid, vi is the radial velocity of the target point i in the road point cloud data, is the radial velocity of the point cloud centroid;
[0431] The point cloud feature extraction module 920 determines the difference between the radial velocity of the target point with the largest value and the radial velocity of the target point with the smallest value in the road point cloud data as the radial velocity change difference of the point cloud mass center, specifically for:
[0432] The radial velocity variation difference of the point cloud mass center is obtained by the following formula:
[0433] a v =v max -v min ;
[0434] Among them, a v is the radial velocity variation difference of the point cloud mass center, v max is the radial velocity of the target point with the largest value in the road point cloud data, v min The radial velocity of the target point with the smallest value in the road point cloud data;
[0435] The point cloud feature extraction module 920 performs the step of determining the mean value of the echo intensity of each target point in the road point cloud data as the point cloud echo intensity mean value, specifically for:
[0436] The mean value of the point cloud echo intensity is obtained by the following formula:
[0437]
[0438] in, is the mean value of the point cloud echo intensity, dB i is the echo intensity of target point i;
[0439] The point cloud feature extraction module 920 performs the step of obtaining the point cloud echo intensity variance according to the echo intensity of each target point in the road point cloud data and the point cloud echo intensity mean, which is specifically used for:
[0440] The point cloud echo intensity variance is obtained by the following formula:
[0441]
[0442] Among them, δ dB is the point cloud echo intensity variance, dB i is the echo intensity of target point i in the road point cloud data, is the mean value of the point cloud echo intensity;
[0443] The point cloud feature extraction module 920 performs the subtraction of the horizontal position coordinate of the target point with the largest value in the road point cloud data from the horizontal position coordinate of the target point with the smallest value to obtain the point cloud horizontal position coordinate difference, which is specifically used for:
[0444] a x =x max -x min ;
[0445] Among them, a x is the horizontal position coordinate difference of the point cloud, x max is the horizontal position coordinate of the target point with the largest value in the road point cloud data, x min is the horizontal position coordinate of the target point with the largest value in the road point cloud data;
[0446] The point cloud feature extraction module 920 performs the subtraction of the longitudinal position coordinate of the target point with the largest value in the road point cloud data from the longitudinal position coordinate of the target point with the smallest value to obtain the point cloud longitudinal position coordinate difference, which is specifically used for:
[0447] The vertical position coordinate difference of the point cloud is obtained by the following formula:
[0448] a y =y max -y min ;
[0449] Among them, a y is the vertical position coordinate difference of the point cloud, y max is the vertical position coordinate of the target point with the largest value in the road point cloud data, y min The vertical position coordinate of the target point with the largest value in the road point cloud data;
[0450] The point cloud feature extraction module 920 performs the step of obtaining the point cloud horizontal position coordinate variance according to the horizontal position coordinates of each target point in the road point cloud data and the horizontal position coordinates of the point cloud centroid, which is specifically used for:
[0451] The horizontal position coordinate variance of the point cloud is obtained by the following formula:
[0452]
[0453] Among them, δ x is the horizontal position coordinate variance of the point cloud, x i is the horizontal position coordinate of the target point i in the road point cloud data, is the horizontal position coordinate of the centroid of the point cloud;
[0454] The point cloud feature extraction module 920 performs the step of obtaining the point cloud longitudinal position coordinate variance according to the longitudinal position coordinates of each target point in the road point cloud data and the longitudinal position coordinates of the point cloud centroid, specifically for:
[0455] The vertical position coordinate variance of the point cloud is obtained by the following formula:
[0456]
[0457] Among them, δ y is the vertical position coordinate variance of the point cloud, y i is the vertical position coordinate of the target point i in the road point cloud data, is the vertical position coordinate of the centroid of the point cloud;
[0458] The point cloud feature extraction module 920 performs the step of obtaining the point cloud density according to the point cloud horizontal position coordinate difference, the point cloud vertical position coordinate difference and the number of point clouds, specifically for:
[0459] The point cloud density is obtained by the following formula:
[0460]
[0461] Where ρ is the point cloud density, α x is the horizontal position coordinate difference of the point cloud, α y is the vertical position coordinate difference of the point cloud.
[0462] In one embodiment, the HRRP of the road point cloud data includes the amplitude of each target point in the road point cloud data; the HRRP geometric features include at least one of the equivalent scattering center number, equivalent target size, HRRP signal entropy, HRRP signal standard deviation, HRRP signal deviation, HRRP signal irregularity, HRRP signal echo energy, HRRP signal energy ratio, HRRP signal skewness and HRRP signal kurtosis; the device also includes:
[0463] The HRRP geometric feature extraction module 960 is used to obtain the equivalent scattering center number by the following formula:
[0464]
[0465] Among them, Num sca is the number of equivalent scattering centers, y(i) is the amplitude of target point i, m is the average amplitude of each target point in the road point cloud data, N is the total number of target points in the road point cloud data, when y(i)-m≥0, ε(y(i)-m)=1, when y(i)-m<0, ε(y(i)-m)=0;
[0466] The equivalent target size is obtained by the following formula:
[0467] Obtaining the equivalent target size according to the position coordinates of the target point with the largest amplitude and the position coordinates of the target point with the smallest amplitude in the HRRP;
[0468] The HRRP signal entropy is obtained by the following formula:
[0469]
[0470] Wherein, Entropy is the HRRP signal entropy;
[0471] The standard deviation of the HRRP signal is obtained by the following formula:
[0472]
[0473] Wherein, Std is the standard deviation of the HRRP signal;
[0474] The HRRP signal deviation is obtained by the following formula:
[0475]
[0476] Deviation is the HRRP signal deviation;
[0477] The HRRP signal irregularity is obtained by the following formula:
[0478]
[0479] Where Irr is the irregularity of the HRRP signal, y(i-1) is the amplitude of the target point i-1 before the target point i, and y(i+1) is the amplitude of the target point i+1 after the target point i;
[0480] The HRRP signal echo energy is obtained by the following formula:
[0481]
[0482] Wherein, TP is the HRRP signal echo energy;
[0483] The HRRP signal energy ratio is obtained by:
[0484] Sorting the amplitudes of the target points in the HRRP from large to small to obtain a sorted HRRP; and dividing the sum of the squares of the amplitudes of the first specified number of target points in the sorted HRRP by the sum of the squares of the amplitudes of the target points in the HRRP to obtain the HRRP signal energy ratio, wherein the specified number is equal to the average value of the amplitudes of the target points;
[0485] The HRRP signal skewness is obtained by the following formula:
[0486]
[0487] Wherein, Skewness is the skewness of the HRRP signal, and μ is the average value of the absolute value of the amplitude of each target point;
[0488] The HRRP signal kurtosis is obtained by the following formula:
[0489]
[0490] Wherein, urtosis is the kurtosis of the HRRP signal.
[0491] In one embodiment, the apparatus further comprises:
[0492] The HRRP power spectrum feature extraction module 970 is used to obtain the HRRP power spectrum features of the road point cloud data by:
[0493] For the amplitude of any target point in the HRRP of the road point cloud data, Fourier transform is performed on the amplitude of the target point to obtain the frequency domain amplitude of the target point; and,
[0494] Determine the square of the frequency domain amplitude of the target point as the power spectrum of the target point;
[0495] The power spectrum of each target point in the HRRP of the road point cloud data is determined as the HRRP power spectrum feature of the road point cloud data.
[0496] After introducing a method and apparatus for multi-target recognition in a complex traffic environment according to an exemplary embodiment of the present disclosure, next, an electronic device according to another exemplary embodiment of the present disclosure is introduced.
[0497] Those skilled in the art will appreciate that various aspects of the present disclosure may be implemented as systems, methods or program products. Therefore, various aspects of the present disclosure may be specifically implemented in the following forms, namely: complete hardware implementation, complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which may be collectively referred to herein as "circuits", "modules" or "systems".
[0498] In some possible implementations, the electronic device according to the present disclosure may include at least one processor and at least one computer storage medium. The computer storage medium stores program code, and when the program code is executed by the processor, the processor executes the steps of the multi-target recognition method in a complex traffic environment according to various exemplary embodiments of the present disclosure described above in this specification. For example, the processor may execute the following steps: Figure 3 Steps 301-305 shown in .
[0499] Refer to the following Fig.10 The electronic device 1000 according to this embodiment of the present disclosure is described. Fig.10 The electronic device 1000 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.
[0500] like Fig.10 As shown, the electronic device 1000 is in the form of a general electronic device. The components of the electronic device 1000 may include but are not limited to: the at least one processor 1001, the at least one computer storage medium 1002, and a bus 1003 connecting different system components (including the computer storage medium 1002 and the processor 1001).
[0501] Bus 1003 represents one or more of several types of bus structures, including a computer storage media bus or computer storage media controller, a peripheral bus, a processor, or a local bus using any of a variety of bus architectures.
[0502] The computer storage medium 1002 may include readable media in the form of volatile computer storage media, such as random access computer storage media (RAM) 1021 and / or cache storage media 1022 , and may further include read-only computer storage media (ROM) 1023 .
[0503] The computer storage medium 1002 may also include a program / utility 1025 having a set (at least one) of program modules 1024, such program modules 1024 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.
[0504] The electronic device 1000 may also communicate with one or more external devices 1004 (e.g., keyboards, pointing devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 1000, and / or communicate with any device that enables the electronic device 1000 to communicate with one or more other electronic devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 1005. Furthermore, the electronic device 1000 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 1006. As shown, the network adapter 1006 communicates with other modules for the electronic device 1000 via a bus 1003. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 1000, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.
[0505] In some possible implementations, various aspects of a method for identifying multiple targets in a complex traffic environment provided by the present disclosure may also be implemented in the form of a program product, which includes program code. When the program product is run on a computer device, the program code is used to enable the computer device to execute the steps of the method for identifying multiple targets in a complex traffic environment according to various exemplary embodiments of the present disclosure described above in this specification.
[0506] The program product may use any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access computer storage medium (RAM), a read-only computer storage medium (ROM), an erasable programmable read-only computer storage medium (EPROM or flash memory), an optical fiber, a portable compact disk read-only computer storage medium (CD-ROM), an optical computer storage medium, a magnetic computer storage medium, or any suitable combination of the above.
[0507] The program product for multi-target recognition in a complex traffic environment of the embodiment of the present disclosure can adopt a portable compact disk read-only computer storage medium (CD-ROM) and include program code, and can be run on an electronic device. However, the program product of the present disclosure is not limited thereto. In this document, a readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, an apparatus or a device.
[0508] The readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, wherein the readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The readable signal medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.
[0509] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.
[0510] Program code for performing the disclosed operations may be written in any combination of one or more programming languages, including object-oriented programming languages such as Java, C++, etc., and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user electronic device, partially on the user device, as a separate software package, partially on the user electronic device and partially on a remote electronic device, or entirely on a remote electronic device or server. In the case of a remote electronic device, the remote electronic device may be connected to the user electronic device via any type of network including a local area network (LAN) or a wide area network (WAN), or may be connected to an external electronic device (e.g., via the Internet using an Internet service provider).
[0511] It should be noted that although several modules of the device are mentioned in the above detailed description, this division is merely exemplary and not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules described above can be embodied in one module. Conversely, the features and functions of one module described above can be further divided into multiple modules to be embodied.
[0512] In addition, although the operations of the disclosed method are described in a specific order in the drawings, this does not require or imply that the operations must be performed in this specific order, or that all the operations shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.
[0513] Those skilled in the art will appreciate that the embodiments of the present disclosure may be provided as methods, systems, or computer program products. Therefore, the present disclosure may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present disclosure may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk computer storage media, CD-ROM, optical computer storage media, etc.) containing computer-usable program codes.
[0514] The present disclosure is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the present disclosure. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0515] These computer program instructions may also be stored in a computer-readable computer storage medium that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable computer storage medium produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0516] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0517] Obviously, those skilled in the art can make various changes and modifications to the present disclosure without departing from the spirit and scope of the present disclosure. Thus, if these modifications and variations of the present disclosure fall within the scope of the claims of the present disclosure and their equivalents, the present disclosure is also intended to include these modifications and variations.
Claims
1. A multi-target recognition method in a complex traffic environment, characterized in that: The method comprises: At specified intervals, obtaining road point cloud data of a specified number of frames, wherein the road point cloud data is obtained by detecting the target road using a millimeter wave radar; For any frame of road point cloud data, project the road point cloud data on the radial distance of the millimeter wave radar to obtain a high-resolution distance signal HRRP of the road point cloud data, and extract point cloud features from the road point cloud data to obtain point cloud features of the road point cloud data; and, Obtaining a point cloud feature set of the road point cloud data according to the point cloud features, HRRP geometric features, and HRRP power spectrum features of the road point cloud data, wherein the HRRP geometric features and the HRRP power spectrum features are obtained by extracting features from the HRRP of the road point cloud data; Obtaining a point cloud feature sequence according to the point cloud feature sets of each of the road point cloud data of the specified number of frames; The point cloud feature sequence is input into a pre-trained target classification model to obtain the category of each target object in the road point cloud data of the specified number of frames.
2. The method according to claim 1, characterized in that The road point cloud data includes the position coordinates, radial velocity and echo intensity of each target point; and any point cloud feature includes at least one of the number of point clouds, the horizontal position coordinate of the point cloud centroid, the vertical position coordinate of the point cloud centroid, the ratio of the horizontal and vertical coordinates of the point cloud centroid, the relative radar distance of the point cloud centroid, the radial velocity of the point cloud centroid, the radial velocity variance of the point cloud centroid, the radial velocity variation difference of the point cloud centroid, the minimum value of the point cloud echo intensity, the maximum value of the point cloud echo intensity, the mean value of the point cloud echo intensity, the variance of the point cloud echo intensity, the horizontal position coordinate difference of the point cloud, the vertical position coordinate difference of the point cloud, the horizontal position coordinate variance of the point cloud, the vertical position coordinate variance of the point cloud and the point cloud density; The step of extracting point cloud features from the road point cloud data to obtain point cloud features of the road point cloud data includes: Determine the total number of target points in the road point cloud data as the point cloud number; and / or, Determine the mean of the horizontal position coordinates of each target point in the road point cloud data as the horizontal position coordinates of the centroid of the point cloud; and / or, Determine the mean value of the longitudinal position coordinates of each target point in the road point cloud data as the longitudinal position coordinates of the centroid of the point cloud; and / or, Dividing the horizontal position coordinate of the point cloud mass center by the vertical position coordinate of the point cloud mass center to obtain the ratio of the horizontal and vertical coordinates of the point cloud mass center; and / or, Obtaining the relative radar distance of the point cloud mass center according to the horizontal position coordinate of the point cloud mass center and the vertical position coordinate of the point cloud mass center; and / or, Determine the average value of the radial velocity of each target point in the road point cloud data as the radial velocity of the point cloud centroid; and / or, According to the radial velocity of each target point in the road point cloud data and the radial velocity of the point cloud centroid, the radial velocity variance of the point cloud centroid is obtained; and / or, Determine the difference between the radial velocity of the target point with the largest value and the radial velocity of the target point with the smallest value in the road point cloud data as the radial velocity change difference of the point cloud mass center; and / or, Determine the echo intensity of the target point with the smallest value in the road point cloud data as the minimum value of the point cloud echo intensity; and / or, Determine the echo intensity of the target point with the largest value in the road point cloud data as the maximum value of the point cloud echo intensity; and / or, Determine the mean value of the echo intensity of each target point in the road point cloud data as the point cloud echo intensity mean value; Obtaining the point cloud echo intensity variance according to the echo intensity of each target point in the road point cloud data and the point cloud echo intensity mean; Subtracting the horizontal position coordinate of the target point with the largest value from the horizontal position coordinate of the target point with the smallest value in the road point cloud data to obtain the horizontal position coordinate difference of the point cloud; Subtracting the longitudinal position coordinate of the target point with the largest value from the longitudinal position coordinate of the target point with the smallest value in the road point cloud data to obtain the longitudinal position coordinate difference of the point cloud; Obtaining the variance of the point cloud horizontal position coordinates according to the horizontal position coordinates of each target point in the road point cloud data and the horizontal position coordinates of the point cloud centroid; Obtaining the variance of the vertical position coordinates of the point cloud according to the vertical position coordinates of each target point in the road point cloud data and the vertical position coordinates of the centroid of the point cloud; The point cloud density is obtained according to the point cloud horizontal position coordinate difference, the point cloud vertical position coordinate difference and the number of point clouds.
3. The method according to claim 2, characterized in that The step of determining the mean of the horizontal position coordinates of each target point in the road point cloud data as the horizontal position coordinates of the centroid of the point cloud includes: The horizontal position coordinates of the centroid of the point cloud are obtained by the following formula: in, is the horizontal position coordinate of the centroid of the point cloud, x i is the horizontal position coordinate of the target point i in the road point cloud data, i∈[1,N], N is the total number of target points in the road point cloud data; The step of determining the mean value of the longitudinal position coordinates of each target point in the road point cloud data as the longitudinal position coordinate of the centroid of the point cloud includes: The vertical position coordinate of the centroid of the point cloud is obtained by the following formula: in, is the vertical position coordinate of the centroid of the point cloud, y i is the vertical position coordinate of the target point i in the road point cloud data; The step of dividing the horizontal position coordinate of the point cloud centroid by the vertical position coordinate of the point cloud centroid to obtain the ratio of the horizontal and vertical coordinates of the point cloud centroid includes: The ratio of the horizontal and vertical coordinates of the point cloud centroid is obtained by the following formula: Wherein, σ is the ratio of the horizontal and vertical coordinates of the centroid of the point cloud; The step of obtaining the relative radar distance of the point cloud mass center according to the horizontal position coordinate of the point cloud mass center and the vertical position coordinate of the point cloud mass center comprises: The relative distance of the point cloud centroid to the radar is obtained by the following formula: Wherein, R is the distance of the point cloud centroid relative to the radar; The step of determining the average value of the radial velocity of each target point in the road point cloud data as the radial velocity of the point cloud centroid includes: The radial velocity of the point cloud centroid is obtained by: in, is the radial velocity of the point cloud centroid, v i is the radial velocity of the target point i in the road point cloud data; The step of obtaining the radial velocity variance of the point cloud centroid according to the radial velocity of each target point in the road point cloud data and the radial velocity of the point cloud centroid comprises: The radial velocity variance of the point cloud centroid is obtained by the following formula: Among them, δ v is the radial velocity variance of the point cloud centroid, v i is the radial velocity of the target point i in the road point cloud data, is the radial velocity of the point cloud centroid; The step of determining the difference between the radial velocity of the target point with the largest value and the radial velocity of the target point with the smallest value in the road point cloud data as the radial velocity change difference of the point cloud mass center includes: The radial velocity variation difference of the point cloud mass center is obtained by the following formula: and v =in max -v min ; Among them, a v is the radial velocity variation difference of the point cloud mass center, v max is the radial velocity of the target point with the largest value in the road point cloud data, v min The radial velocity of the target point with the smallest value in the road point cloud data; The step of determining the mean value of the echo intensity of each target point in the road point cloud data as the point cloud echo intensity mean value comprises: The mean value of the point cloud echo intensity is obtained by the following formula: in, is the mean value of the point cloud echo intensity, dB i is the echo intensity of target point i; The step of obtaining the point cloud echo intensity variance according to the echo intensity of each target point in the road point cloud data and the point cloud echo intensity mean value includes: The point cloud echo intensity variance is obtained by the following formula: Among them, δ dB is the point cloud echo intensity variance, dB i is the echo intensity of target point i in the road point cloud data, is the mean value of the point cloud echo intensity; Subtracting the horizontal position coordinate of the target point with the largest value from the horizontal position coordinate of the target point with the smallest value in the road point cloud data to obtain the horizontal position coordinate difference of the point cloud includes: a x =x max -x min ; Among them, a x is the horizontal position coordinate difference of the point cloud, x max is the horizontal position coordinate of the target point with the largest value in the road point cloud data, x min is the horizontal position coordinate of the target point with the largest value in the road point cloud data; Subtracting the longitudinal position coordinate of the target point with the largest value from the longitudinal position coordinate of the target point with the smallest value in the road point cloud data to obtain the longitudinal position coordinate difference of the point cloud includes: The vertical position coordinate difference of the point cloud is obtained by the following formula: to y =and max -and min ; Among them, a y is the vertical position coordinate difference of the point cloud, y max is the vertical position coordinate of the target point with the largest value in the road point cloud data, y min The vertical position coordinate of the target point with the largest value in the road point cloud data; The step of obtaining the variance of the point cloud horizontal position coordinates according to the horizontal position coordinates of each target point in the road point cloud data and the horizontal position coordinates of the point cloud centroid comprises: The horizontal position coordinate variance of the point cloud is obtained by the following formula: Among them, δ x is the horizontal position coordinate variance of the point cloud, x i is the horizontal position coordinate of the target point i in the road point cloud data, is the horizontal position coordinate of the centroid of the point cloud; The method of obtaining the point cloud longitudinal position coordinate variance according to the longitudinal position coordinate of each target point in the road point cloud data and the longitudinal position coordinate of the point cloud centroid comprises: The vertical position coordinate variance of the point cloud is obtained by the following formula: Among them, δ y is the vertical position coordinate variance of the point cloud, y i is the vertical position coordinate of the target point i in the road point cloud data, is the vertical position coordinate of the centroid of the point cloud; The step of obtaining the point cloud density according to the horizontal position coordinate difference of the point cloud, the vertical position coordinate difference of the point cloud and the number of point clouds includes: The point cloud density is obtained by the following formula: Where ρ is the point cloud density, α x is the horizontal position coordinate difference of the point cloud, α y is the vertical position coordinate difference of the point cloud.
4. The method according to claim 1, characterized in that: The HRRP of the road point cloud data includes the amplitude of each target point in the road point cloud data; The HRRP geometric features include at least one of the equivalent scattering center number, equivalent target size, HRRP signal entropy, HRRP signal standard deviation, HRRP signal deviation, HRRP signal irregularity, HRRP signal echo energy, HRRP signal energy ratio, HRRP signal skewness and HRRP signal kurtosis; The equivalent scattering center number is obtained by the following formula: Among them, Num sca is the number of equivalent scattering centers, y(i) is the amplitude of target point i, m is the average amplitude of each target point in the road point cloud data, N is the total number of target points in the road point cloud data, when y(i)-m≥0, ε(y(i)-m)=1, when y(i)-m<0, ε(y(i)-m)=0; The equivalent target size is obtained by the following formula: Obtaining the equivalent target size according to the position coordinates of the target point with the largest amplitude and the position coordinates of the target point with the smallest amplitude in the HRRP; The HRRP signal entropy is obtained by the following formula: Wherein, Entropy is the HRRP signal entropy; The standard deviation of the HRRP signal is obtained by the following formula: Wherein, Std is the standard deviation of the HRRP signal; The HRRP signal deviation is obtained by the following formula: Deviation is the HRRP signal deviation; The HRRP signal irregularity is obtained by the following formula: Where Irr is the irregularity of the HRRP signal, y(i-1) is the amplitude of the target point i-1 before the target point i, and y(i+1) is the amplitude of the target point i+1 after the target point i; The HRRP signal echo energy is obtained by the following formula: Wherein, TP is the HRRP signal echo energy; The HRRP signal energy ratio is obtained by: Sorting the amplitudes of the target points in the HRRP from large to small to obtain a sorted HRRP; and dividing the sum of the squares of the amplitudes of the first specified number of target points in the sorted HRRP by the sum of the squares of the amplitudes of the target points in the HRRP to obtain the HRRP signal energy ratio, wherein the specified number is equal to the average value of the amplitudes of the target points; The HRRP signal skewness is obtained by the following formula: Wherein, Skewness is the skewness of the HRRP signal, and μ is the average value of the absolute value of the amplitude of each target point; The HRRP signal kurtosis is obtained by the following formula: Wherein, urtosis is the kurtosis of the HRRP signal.
5. The method according to claim 1, characterized in that The HRRP power spectrum characteristics of the road point cloud data are obtained by: For the amplitude of any target point in the HRRP of the road point cloud data, Fourier transform is performed on the amplitude of the target point to obtain the frequency domain amplitude of the target point; and, Determine the square of the frequency domain amplitude of the target point as the power spectrum of the target point; The power spectrum of each target point in the HRRP of the road point cloud data is determined as the HRRP power spectrum feature of the road point cloud data.
6. An electronic device, characterized in that: comprising a processor and a memory, wherein the processor and the memory are connected via a bus; The memory stores a computer program, and the processor is configured to perform the following operations based on the computer program: At specified intervals, obtaining road point cloud data of a specified number of frames, wherein the road point cloud data is obtained by detecting the target road using a millimeter wave radar; For any frame of road point cloud data, project the road point cloud data on the radial distance of the millimeter wave radar to obtain a high-resolution distance signal HRRP of the road point cloud data, and extract point cloud features from the road point cloud data to obtain point cloud features of each of the road point cloud data; and, Obtaining a point cloud feature set of the road point cloud data according to the point cloud features, HRRP geometric features, and HRRP power spectrum features of the road point cloud data, wherein the HRRP geometric features and the HRRP power spectrum features are obtained by extracting features from the HRRP of the road point cloud data; Obtaining a point cloud feature sequence according to the point cloud feature set of the road point cloud data of the specified number of frames; The point cloud feature sequence is input into a pre-trained target classification model to obtain the category of each target object in the road point cloud data of the specified number of frames.
7. The electronic device according to claim 6, characterized in that: The road point cloud data includes the position coordinates, radial velocity and echo intensity of each target point; and any point cloud feature includes at least one of the number of point clouds, the horizontal position coordinate of the point cloud centroid, the vertical position coordinate of the point cloud centroid, the ratio of the horizontal and vertical coordinates of the point cloud centroid, the relative radar distance of the point cloud centroid, the radial velocity of the point cloud centroid, the radial velocity variance of the point cloud centroid, the radial velocity variation difference of the point cloud centroid, the minimum value of the point cloud echo intensity, the maximum value of the point cloud echo intensity, the mean value of the point cloud echo intensity, the variance of the point cloud echo intensity, the horizontal position coordinate difference of the point cloud, the vertical position coordinate difference of the point cloud, the horizontal position coordinate variance of the point cloud, the vertical position coordinate variance of the point cloud and the point cloud density; The processor performs the point cloud feature extraction on the road point cloud data to obtain the point cloud features of the road point cloud data, and is specifically configured as follows: Determine the total number of target points in the road point cloud data as the point cloud number; and / or, Determine the mean of the horizontal position coordinates of each target point in the road point cloud data as the horizontal position coordinates of the centroid of the point cloud; and / or, Determine the mean value of the longitudinal position coordinates of each target point in the road point cloud data as the longitudinal position coordinates of the centroid of the point cloud; and / or, Dividing the horizontal position coordinate of the point cloud mass center by the vertical position coordinate of the point cloud mass center to obtain the ratio of the horizontal and vertical coordinates of the point cloud mass center; and / or, Obtaining the relative radar distance of the point cloud mass center according to the horizontal position coordinate of the point cloud mass center and the vertical position coordinate of the point cloud mass center; and / or, Determine the average value of the radial velocity of each target point in the road point cloud data as the radial velocity of the point cloud centroid; and / or, According to the radial velocity of each target point in the road point cloud data and the radial velocity of the point cloud centroid, the radial velocity variance of the point cloud centroid is obtained; and / or, Determine the difference between the radial velocity of the target point with the largest value and the radial velocity of the target point with the smallest value in the road point cloud data as the radial velocity change difference of the point cloud mass center; and / or, Determine the echo intensity of the target point with the smallest value in the road point cloud data as the minimum value of the point cloud echo intensity; and / or, Determine the echo intensity of the target point with the largest value in the road point cloud data as the maximum value of the point cloud echo intensity; and / or, Determine the mean value of the echo intensity of each target point in the road point cloud data as the point cloud echo intensity mean value; Obtaining the point cloud echo intensity variance according to the echo intensity of each target point in the road point cloud data and the point cloud echo intensity mean; Subtracting the horizontal position coordinate of the target point with the largest value from the horizontal position coordinate of the target point with the smallest value in the road point cloud data to obtain the horizontal position coordinate difference of the point cloud; Subtracting the longitudinal position coordinate of the target point with the largest value from the longitudinal position coordinate of the target point with the smallest value in the road point cloud data to obtain the longitudinal position coordinate difference of the point cloud; Obtaining the variance of the point cloud horizontal position coordinates according to the horizontal position coordinates of each target point in the road point cloud data and the horizontal position coordinates of the point cloud centroid; Obtaining the variance of the vertical position coordinates of the point cloud according to the vertical position coordinates of each target point in the road point cloud data and the vertical position coordinates of the centroid of the point cloud; The point cloud density is obtained according to the point cloud horizontal position coordinate difference, the point cloud vertical position coordinate difference and the number of point clouds.
8. The electronic device according to claim 7, characterized in that: The processor determines the mean of the horizontal position coordinates of each target point in the road point cloud data as the horizontal position coordinates of the centroid of the point cloud, and is specifically configured as follows: The horizontal position coordinates of the centroid of the point cloud are obtained by the following formula: in, is the horizontal position coordinate of the centroid of the point cloud, x i is the horizontal position coordinate of the target point i in the road point cloud data, i∈[1,N], N is the total number of target points in the road point cloud data; The processor performs the step of determining the mean value of the longitudinal position coordinates of each target point in the road point cloud data as the longitudinal position coordinates of the centroid of the point cloud, and is specifically configured as follows: The vertical position coordinate of the centroid of the point cloud is obtained by the following formula: in, is the vertical position coordinate of the centroid of the point cloud, y i is the vertical position coordinate of the target point i in the road point cloud data; The processor performs the division of the horizontal position coordinate of the point cloud centroid by the vertical position coordinate of the point cloud centroid to obtain the ratio of the horizontal and vertical coordinates of the point cloud centroid, and is specifically configured as follows: The ratio of the horizontal and vertical coordinates of the point cloud centroid is obtained by the following formula: Wherein, σ is the ratio of the horizontal and vertical coordinates of the centroid of the point cloud; The processor executes the step of obtaining the relative radar distance of the point cloud centroid according to the horizontal position coordinate of the point cloud centroid and the vertical position coordinate of the point cloud centroid, and is specifically configured as follows: The relative distance of the point cloud centroid to the radar is obtained by the following formula: Wherein, R is the distance of the point cloud centroid relative to the radar; The processor performs the step of determining the average value of the radial velocity of each target point in the road point cloud data as the radial velocity of the point cloud centroid, and is specifically configured as follows: The radial velocity of the point cloud centroid is obtained by: in, is the radial velocity of the point cloud centroid, v i is the radial velocity of the target point i in the road point cloud data; The processor executes the step of obtaining the radial velocity variance of the point cloud centroid according to the radial velocity of each target point in the road point cloud data and the radial velocity of the point cloud centroid, and is specifically configured as follows: The radial velocity variance of the point cloud centroid is obtained by the following formula: Among them, δ v is the radial velocity variance of the point cloud centroid, v i is the radial velocity of the target point i in the road point cloud data, is the radial velocity of the point cloud centroid; The processor performs the step of determining the difference between the radial velocity of the target point with the largest value and the radial velocity of the target point with the smallest value in the road point cloud data as the radial velocity change difference of the point cloud mass center, and is specifically configured as follows: The radial velocity variation difference of the point cloud mass center is obtained by the following formula: and v =in max -v min ; Among them, a v is the radial velocity variation difference of the point cloud mass center, v max is the radial velocity of the target point with the largest value in the road point cloud data, v min The radial velocity of the target point with the smallest value in the road point cloud data; The processor performs the step of determining the mean value of the echo intensity of each target point in the road point cloud data as the point cloud echo intensity mean value, and is specifically configured as follows: The mean value of the point cloud echo intensity is obtained by the following formula: in, is the mean value of the point cloud echo intensity, dB i is the echo intensity of target point i; The processor executes the step of obtaining the point cloud echo intensity variance according to the echo intensity of each target point in the road point cloud data and the point cloud echo intensity mean, and is specifically configured as follows: The point cloud echo intensity variance is obtained by the following formula: Among them, δ dB is the point cloud echo intensity variance, dB i is the echo intensity of target point i in the road point cloud data, is the mean value of the point cloud echo intensity; The processor performs the step of subtracting the horizontal position coordinate of the target point with the largest value from the horizontal position coordinate of the target point with the smallest value in the road point cloud data to obtain the horizontal position coordinate difference of the point cloud, and is specifically configured as follows: a x =x max -x min ; Among them, a x is the horizontal position coordinate difference of the point cloud, x max is the horizontal position coordinate of the target point with the largest value in the road point cloud data, x min is the horizontal position coordinate of the target point with the largest value in the road point cloud data; The processor performs the subtraction of the longitudinal position coordinate of the target point with the largest value in the road point cloud data from the longitudinal position coordinate of the target point with the smallest value to obtain the longitudinal position coordinate difference of the point cloud, and is specifically configured as follows: The vertical position coordinate difference of the point cloud is obtained by the following formula: to y =and max -and min ; Among them, a y is the vertical position coordinate difference of the point cloud, y max is the vertical position coordinate of the target point with the largest value in the road point cloud data, y min The vertical position coordinate of the target point with the largest value in the road point cloud data; The processor executes the step of obtaining the variance of the point cloud horizontal position coordinates according to the horizontal position coordinates of each target point in the road point cloud data and the horizontal position coordinates of the point cloud centroid, and is specifically configured as follows: The horizontal position coordinate variance of the point cloud is obtained by the following formula: Among them, δ x is the horizontal position coordinate variance of the point cloud, x i is the horizontal position coordinate of the target point i in the road point cloud data, is the horizontal position coordinate of the centroid of the point cloud; The processor executes the step of obtaining the variance of the vertical position coordinates of the point cloud according to the vertical position coordinates of each target point in the road point cloud data and the vertical position coordinates of the centroid of the point cloud, and is specifically configured as follows: The vertical position coordinate variance of the point cloud is obtained by the following formula: Among them, δ y is the vertical position coordinate variance of the point cloud, y i is the vertical position coordinate of the target point i in the road point cloud data, is the vertical position coordinate of the centroid of the point cloud; The processor executes the step of obtaining the point cloud density according to the point cloud horizontal position coordinate difference, the point cloud vertical position coordinate difference and the number of point clouds, and is specifically configured as follows: The point cloud density is obtained by the following formula: Where ρ is the point cloud density, α x is the horizontal position coordinate difference of the point cloud, α y is the vertical position coordinate difference of the point cloud.
9. The electronic device according to claim 6, characterized in that: The HRRP of the road point cloud data includes the amplitude of each target point in the road point cloud data; The HRRP geometric features include at least one of the equivalent scattering center number, equivalent target size, HRRP signal entropy, HRRP signal standard deviation, HRRP signal deviation, HRRP signal irregularity, HRRP signal echo energy, HRRP signal energy ratio, HRRP signal skewness and HRRP signal kurtosis; The processor is further configured to: The equivalent scattering center number is obtained by the following formula: Among them, Num sca is the number of equivalent scattering centers, y(i) is the amplitude of target point i, m is the average amplitude of each target point in the road point cloud data, N is the total number of target points in the road point cloud data, when y(i)-m≥0, ε(y(i)-m)=1, when y(i)-m<0, ε(y(i)-m)=0; The equivalent target size is obtained by the following formula: Obtaining the equivalent target size according to the position coordinates of the target point with the largest amplitude and the position coordinates of the target point with the smallest amplitude in the HRRP; The HRRP signal entropy is obtained by the following formula: Wherein, Entropy is the HRRP signal entropy; The standard deviation of the HRRP signal is obtained by the following formula: Wherein, Std is the standard deviation of the HRRP signal; The HRRP signal deviation is obtained by the following formula: Deviation is the HRRP signal deviation; The HRRP signal irregularity is obtained by the following formula: Where Irr is the irregularity of the HRRP signal, y(i-1) is the amplitude of the target point i-1 before the target point i, and y(i+1) is the amplitude of the target point i+1 after the target point i; The HRRP signal echo energy is obtained by the following formula: Wherein, TP is the HRRP signal echo energy; The HRRP signal energy ratio is obtained by: Sorting the amplitudes of the target points in the HRRP from large to small to obtain a sorted HRRP; and dividing the sum of the squares of the amplitudes of the first specified number of target points in the sorted HRRP by the sum of the squares of the amplitudes of the target points in the HRRP to obtain the HRRP signal energy ratio, wherein the specified number is equal to the average value of the amplitudes of the target points; The HRRP signal skewness is obtained by the following formula: Wherein, Skewness is the skewness of the HRRP signal, and μ is the average value of the absolute value of the amplitude of each target point; The HRRP signal kurtosis is obtained by the following formula: Wherein, urtosis is the kurtosis of the HRRP signal.
10. The electronic device according to claim 6, characterized in that: The processor is further configured to: The HRRP power spectrum characteristics of the road point cloud data are obtained by: For the amplitude of any target point in the HRRP of the road point cloud data, Fourier transform is performed on the amplitude of the target point to obtain the frequency domain amplitude of the target point; and, Determine the square of the frequency domain amplitude of the target point as the power spectrum of the target point; The power spectrum of each target point in the HRRP of the road point cloud data is determined as the HRRP power spectrum feature of the road point cloud data.
Citation Information
Patent Citations
Method and device for acquiring laser imaging echo waveform and level characteristics
CN101839981A
Video image target positioning method based on three-dimensional laser point cloud
CN110619663A
Super-resolution processing method, device and system for laser point cloud data and storage medium
CN111127315A
High-resolution processing method and device and computer storage medium
CN114202461A
Apparatus and method for 3D depth-data upsampling
KR102482532B1
