Hyundai Motor Company

Kia Corporation

Yonsei University

Eun Tai Kim

Hyun Ju Kim

Jin Kyu Hwang

Won Je Jang

Min Seong Park

Jong Won Park

So Jin Jang

A method of classifying an object according to an embodiment includes extracting a first feature by transforming rectangular coordinates of points included in the box of the object, obtained from a point cloud acquired using a LiDAR sensor, into complex coordinates and performing Fast Fourier Transform (FFT) on the complex coordinates, obtaining an average and a standard deviation as a second feature, the average and the standard deviation being parameters of a Gaussian model for the points included in the box of the object, and classifying the type of object based on at least one of the first feature or the second feature.

Method and apparatus for classifying object and recording medium storing program to execute the method

An embodiment device for detecting a boundary of a road in a 3D point cloud using a cascade classifier includes a rule-based classifier configured to determine whether a received LiDAR cluster has a likelihood of becoming a candidate for the boundary of the road using a box parameter surrounding a point cloud constituting the LiDAR cluster and a point parameter, and a learning-based classifier configured to apply a machine-learning scheme to the LiDAR cluster selected as the candidate for the boundary of the road by the rule-based classifier to determine the LiDAR cluster to be the boundary of the road or an object other than the boundary of the road.

Method and device for detecting boundary of road in 3D point cloud using cascade classifier

Growing community of inventors

Seoul, South Korea

Eun Tai Kim