Multi-target classification method based on vehicle-mounted millimeter-wave radar combined with svm and cnn

Abstract

The invention discloses a multi-target classification method based on vehicle-mounted millimeter-wave radar combined with SVM and CNN. First, the IF signals of different targets acquired by the radar are sampled, converted into frame signals, and two-dimensional Fourier transform is performed on the frame signals. After normalization, the distance-Doppler map is obtained, the feature vector in the distance-Doppler map is extracted, and the sample set is constructed; secondly, the SVM classifier is constructed and trained, and the SVM classifier capable of preliminary classification is obtained; finally, it is constructed and trained The convolutional neural network classifier further classifies the samples that the SVM classifier cannot classify, and obtains the classification result of the target to be tested. This method combines the SVM classifier and the CNN classifier to make up for the insensitivity of the CNN classifier to the position information, make full use of the effective information of the samples, and improve the accuracy of the target classification.

Description

technical field [0001] The invention relates to a target classification method based on a vehicle-mounted millimeter-wave sensor, in particular to a target classification method based on a vehicle-mounted millimeter-wave sensor combined support vector machine and convolutional 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. More and more enterprises and scientific research institutions have actively participated in and promoted the development of unmanned driving. Since the automotive industry has extremely high requirements for pedestrian safety, pedestrian and vehicle classification has gradually become a key technology in driverless driving. In the field of autonomous driving, unmanned vehicles must have the ability to identify pedestrians and vehicles and their l...

AI Technical Summary

Problems solved by technology

Convolutional neural networks are sensitive to training set sample imbalance, and the "more imbalanced" the category, the worse the classification effect

At the same time, because the convolution kernel is used to extract the local features of the image in the convolutional neural network, the position information of the target itself is ignored during the classifier training process, and these features can also be used for target classification in some cases.

Images