A multi-target classification method based on vehicle-mounted millimeter-wave radar

A technology of millimeter-wave radar and classification method, which is applied in neural learning methods, instruments, biological neural network models, etc., can solve the problems of limited application range of cascaded classifiers and cannot be used to solve multi-target classification of samples, so as to improve accuracy efficiency, the effect of overcoming limitations

Active Publication Date: 2021-03-26
SOUTHEAST UNIV
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, cascaded classifiers have limited application range and cannot be used to solve multi-objective classification with sample imbalance.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A multi-target classification method based on vehicle-mounted millimeter-wave radar
  • A multi-target classification method based on vehicle-mounted millimeter-wave radar
  • A multi-target classification method based on vehicle-mounted millimeter-wave radar

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0062] like figure 1 As shown, a multi-target classification method based on a vehicle-mounted millimeter-wave radar includes: a training phase and a testing phase, and the training phase includes the following steps:

[0063] Step 1. Obtain the intermediate frequency signal f(t) processed by the millimeter wave radar system on the target echo signal, and demarcate the classification label for each obtained intermediate frequency signal f(t);

[0064] The millimeter-wave radar system is installed on a vehicle, and the millimeter-wave radar system includes a radar transmitter, a radar receiver, and a mixer; the radar transmitter periodically transmits a chirp signal, and the radar receiver receives the echoes scattered by the target. wave signal, the mixer uses the linear frequency modulation signal transmitted by the radar to mix the received echo signal to obtain the intermediate frequency signal f(t);

[0065] The structure of the vehicle millimeter-wave radar system is as ...

Embodiment 2

[0101] Embodiment 1 constructs a sample set by acquiring the intermediate frequency signal f(t) after the millimeter wave radar system processes the target echo signal. When the collected intermediate frequency signal f(t) of the known target category is insufficient, the intermediate frequency signal f(t) can be generated by the simulation method, including the following steps:

[0102] (1.1) Establish the time-domain radar echo signal expressions of pedestrians, bicycles and cars:

[0103]

[0104] where M is the number of scattering points of the target, ρ k is the scattering coefficient of the kth scattering point, τ k =2R k / c is the echo delay of the kth scattering point, R k Represents the distance between the radar and the kth scattering point, c is the propagation speed of electromagnetic waves; f c is the carrier frequency of the chirp signal transmitted by the radar, and γ is the chirp slope of the chirp signal transmitted by the radar;

[0105] By modeling ...

Embodiment 3

[0142] In this embodiment, 40 IF signals of pedestrians, 80 IF signals of bicycles, and 200 IF signals of automobiles are collected, and each collected IF signal is processed to generate multiple range-Doppler maps. From the multiple range-Doppler maps generated by the signal, 5 range-Doppler maps are extracted at equal intervals to form a sample set, with a total of 1600 maps. In the sample set, 400 distance-Doppler images were selected to form the validation set, and the remaining 1200 images were used as the training set. In addition, a set of 400 test samples was generated in the same way.

[0143] This example builds a hybrid cascaded neural network classifier on the Tensorflow framework and uses GPU for accelerated training. The size of the sample set data image is 224*224, and the size of the image is converted to 56*56 through linear interpolation, and the corresponding mean value is subtracted from the gray value of each pixel point, and then sent to the network for ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a multi-target classification method based on a vehicle-mounted millimeter-wave radar. Firstly, the IF signals of different targets acquired by the radar are sampled, converted into frame signals, and the frame signals are subjected to two-dimensional Fourier transform and normalized. Obtain the distance-Doppler map afterward; Construct the distance-Doppler map sample set thus; Secondly construct the hybrid cascade neural network classifier, use the distance-Doppler map sample set as input to the hybrid cascade neural network classifier Supervised learning is carried out to obtain the network parameters of the classifiers on each branch, and finally a classifier capable of multi-target classification is obtained to classify the target IF signals acquired by the radar. This method overcomes the limitations of cascade classifier application scenarios by mixing cascade classifiers, and can classify multiple target categories.

Description

technical field [0001] The invention relates to a multi-target classification method based on a vehicle-mounted millimeter-wave radar, in particular to a vehicle-mounted millimeter-wave radar multi-target classification method based on a hybrid cascaded neural network. Background technique [0002] In recent years, with the continuous improvement of the market's demand for active safety and intelligence of automobiles, the huge social and economic value of unmanned driving has become more and more prominent, and more and more enterprises and scientific research institutions are actively participating in and promoting the development of unmanned driving. Since the automotive industry has extremely high requirements for the safety of pedestrians, classification of pedestrians and vehicles has gradually become a key technology in autonomous driving. In the field of autonomous driving, driverless vehicles must have the ability to identify pedestrians and vehicles and their locat...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & AuthorityPatents(China)
IPC IPC(8): G06K9/00G06N3/04G06N3/08
CPCG06N3/08G06N3/044G06N3/045G06F2218/12
Inventor武其松高腾
OwnerSOUTHEAST UNIV