A Novel Multi-Geomagnetic Vehicle Sensors Data Feature Alignment Path Search Method of DTW
He nana, Zheng xuueyan · Research Square · 2022
Abstract The recognition of vehicle information based on vehicle geomagnetic information has great advantages in ITS. Traditionally, when using a single geomagnetic sensor to detect vehicle information, less vehicle information will be obtained, and will produce large classification errors when classifying the vehicle types. After pre-processing, vehicle geomagnetic data extraction, data fusion, feature extraction and vehicle type classification, real-time vehicle information can be obtained by using multiple geomagnetic sensors for detection. In this paper, the feature points in the vehicle geomagnetic data collected by multiple sensors are aligned based on the DTW method, and the method of path search which is most suitable for the alignment of feature points of vehicle geomagnetic data is found out from many method of path search of DTW. Experiments show that compared with other methods of path search, this method has the advantages of better preservation of original data, better alignment of feature points and less computation.