The invention discloses a
vehicle detection and identification method and
system based on
laser sensing, and relates to the field of
vehicle detection.The method comprises the steps that firstly, obstacle reflection signals are obtained through a multi-band
laser array, preliminary classification is conducted through
wavelength,
time difference and intensity difference, and
point cloud features are constructed; a lightweight network is adopted to
filter noise points, and
point cloud holes and shielding areas are intelligently complemented; then, performing hierarchical voxelization
processing according to the obstacle distance, performing coarse-grained downsampling on a long-distance target, and keeping details on a short-distance target fine-grained; and finally, through a double-stage
verification mechanism of a lightweight PointPill network and a three-dimensional convolutional network, precise grading detection and target reliability rechecking of far and near targets are realized, real-time performance and accuracy are considered, the problems of poor adaptability, low precision and insufficient dynamic scene adaptability of an existing
laser radar detection technology are solved, and the method has the advantages of high detection precision and high reliability. According to the invention, high-precision and high-real-time
vehicle detection and identification can be realized.