Asier Izquierdo Pérez. "Intelligent road lane mark extraction using a Mobile Mapping System."

Zuzendariak_Directores: Manuel Maria Graña Romay / Jose Manuel  Lopez Guede .

2023_01_19, 10:00  Sala Ada Lovelace aretoa.

Abstract:

"This Thesis deals with the task of detection of different road lane marks, us­ing signals from the different sensors that a Mobile Mapping System (MMS) is composed of, namely image sensors (cameras) and LiDAR sensor. MMSs are the combination of various navigation and remote sensing technologies on a common moving platform. During the last years, road landmark inventory has raised increasing interest in different areas: the maintenance of transport infrastructures, road 3d modelling, GIS applications, etc. Several commercial sensors are available which include a set of high-resolution cameras already calibrated in order to generate panoramic images, and a LiDAR sensor that al­lows to capture up to 700,000 georeferenced points, plus other components that provide ancillary information. The lane mark detection is posed as a two-class classification problem over a highly class imbalanced dataset. To cope with this imbalance we have applied Active Learning approaches. This Thesis has been divided into two main computational parts. In the first part, we have evaluated different Machine Learning approaches using panoramic images, obtained from image sensor, such as Random Forest (R.F) and ensembles of Extreme Learn­ing Machines (V-ELM), obtaining satisfactory results in the detection of road continuous lane marks. In the second part of the Thesis, we have applied a Ran­dom Forest algorithm to a LiDAR point cloud, obtaining a georeferenced road horizontal signs classification. We have not only identified continuous lines, but also, we have heen able to identify every horizontal lane mark detected by the LiDAR sensor. 
Keywords Mobile Mapping System, Active Learning, Random Forest, Extreme Learning Machine, LiDAR, Image Sensor, Panoramic image ".