Road Surface Classification Based on Radar Imaging Using Convolutional Neural Network
Shahrzad Minooee Sabery, Aleksandr Bystrov, P. Gardner, Ana Stroescu, Marina Gashinova
Abstract
The development of an automotive surface recognition system is an important and yet unsolved task. In the current study we are considering a novel approach to surface classification based on the analysis of the real road surface images obtained using the 79 GHz imaging radar and demonstrate the advantage of millimeter wave radar for surface discrimination for automotive sensing. The proposed experimental technique in combination with a convolutional neural network provides high surface classification accuracy.
Topics & Concepts
Convolutional neural networkRadarRadar imagingAutomotive industryComputer scienceArtificial intelligenceArtificial neural networkSurface (topology)Extremely high frequencyPattern recognition (psychology)Road surfaceComputer visionRemote sensingEngineeringGeologyTelecommunicationsAerospace engineeringMathematicsCivil engineeringGeometryAdvanced Optical Sensing TechnologiesRemote Sensing and LiDAR ApplicationsGeophysical Methods and Applications