IQmulus Processing Contest 2013

An initiative similar to the IQmulus processing contest has started in 2011 under the aegis of the ISPRS (Rottensteiner et al., 2012). The need of new standard test sites that exploit the availability of modern airborne sensors led to the creation of a benchmark on urban object extraction. In particular, both airborne topographic Lidar data and multi‐view stereo imagery are provided to compare and evaluate research algorithms for building detection, tree detectionand building reconstruction in urban areas. In addition, a benchmark dataset on shape registration has been recently proposed in the field of robotic research (Pomerleau et al.,2013). The results of the contest highlighted that there is still place for further developments in all the scenarios considered; in particular the detection of small building structures and trees and the production of high‐quality level‐of‐detail building models. As a further outcome, the methods currently at the state of the art do not seem to be able to fully exploit the accuracy potential inherent in the sensor data.

IQPC '13 Tracks

1. The registration of sets of point clouds

Six datasets have been selected to evaluate the properties of the methods and are provided to the participants. The datasets answer to problems that arise in data registration, for instance the registration of point clouds acquired in two different epochs (for instance the Amiens_AirborneLidar dataset); the registration of crowded urban streets acquired with a mobile vehicle (for instance the datasets Paris1_MMS and Paris3_MMS); the registration of different airborne LiDAR (for instance the datasets originating from Regione Liguria data).

2. Feature extraction (in particular the extraction of drainage basins from a DEM)

Three datasets are proposed for the track on the extraction of drainage basins. The three models come from the DEM of the Liguria area, as provided by Regione Liguria. The DEM models are in geotiff format and correspond to a sampling of the area with a pass of 5 metres. The algorithms in the literature for the extraction of medium‐large drainage basins (more than 10mq) perform very well and reliably. The challenge posed by these datasets is the selection of medium/small drainage basins.


Organization

IQPC’13 Organizers: Silvia Biasotti, IMATI-CNR, Italy

IQPC’13 Committee/Advisory Board: Jan Boehm, UCL, UK, - Clément Mallet, IGN, France - Roderik Lindenbergh, TUDelft, NL, -Silvia Biasotti, IMATI-CNR, Italy


The Report on the IQmulus Processing Contest Year 1 available here.