Vehicle type classification from laser scans with global alignment kernels

This paper addresses the problem of vehicle classification using laser scanner profiles, which is usually found as a component of electronic tolling systems. Laser scanners obtain a 3D measurement of the vehicle surface. Previous approaches treated the laser scans as images. In addition to high-level features (such as width, height, length and other measurements) from the scanner profiles, feature descriptors have been used for supervised classification of laser scanner profiles. This allowed to deploy a generic approach to the image classification based on Fisher features. In this paper we adopt a different approach which extracts and analyses vehicle shapes from the laser scans. The method proceeds to finding typical vehicle shapes and classifying them through the global shape alignment.