ISPRS Geospatial Week Recap

Thanks to Dr. Michael Olsen  at Oregon State University for the following recap of last week’s ISPRS Geospatial week held in France. Some guys have all the luck.

Bonjour!  This week I had the privilege of attending the ISPRS Geospatial Week held in La Grande Motte, France and will provide you with a short overview.  Basically, the ISPRS Geospatial Week combines several working groups/workshops within ISPRS including LaserScanning and SilviLaser that have traditionally held individual workshops.  I predominately attended the laser scanning workshop, but heard great things from other attendees about all of the other events going on this week so far.

The LaserScanning Working Group has had biannual meetings since 1999 (although it had a different name until 2005).  The papers presented here are of very high quality and on the cutting edge.  This conference primarily consists of scientists from across the globe who are developing state of the art algorithms for lidar processing (e.g., registration, feature extraction, classification) and/or are involved in new techniques for calibration.  It is a great opportunity to rub shoulders with many of the leading experts in these areas.  While the conference contains presentations by those primarily in the academic sector, I ran into several public or private sector professionals who were very impressed by the content.

Peer Reviewed Papers submitted to the conference can be found at the ISPRS Annals located at:

Many of you have probably come across a publication from a previous laser scanning workshop.  I have used them frequently in literature reviews for research.  The workshop kicked off with some plenary and keynote speakers, including Mike Wulder and Francis Halle who both spoke about scientific insights that are provided by lidar for forest ecosystems and potential opportunities for future research.

Some highlights of the other sessions of the conference include:

  • New, automated techniques for point cloud registration, including some novel ones that account for the uncertainty of individual data points. A key challenge in point cloud registration is that many algorithms require a good, initial alignment, which can be a very tedious process. These new algorithms presented show great promise in reducing the manual work to obtain a coarse alignment that can be refined with the existing techniques. Others approaches utilize computer vision algorithms to extract matching point pairs from panoramic image maps of lidar intensity.
  • An approach to improve the accuracy of mobile lidar data by acquiring multiple passes and using the redundancy to remove outliers caused by GNSS Multipath. (I’m a little biased on this one since I was a co-author of this study).
  • Several other researchers developed new techniques for rapid building reconstruction from point cloud datasets. Some were developed for a single building while others function across an entire city. These techniques worked well at preserving features in the dataset while removing redundant data.
  • Segmentation and classification of point clouds including new algorithms to rapidly extract planes, poles, and trees. (There were quite a few papers on individual tree extraction). Researchers at UC Calgary have developed a new approach to construction progress monitoring based on point cloud segmentation approaches and comparison to as-built geometries. Another paper focused on extracting individual bricks automatically using a clever approach using intensity images of the point cloud. They apply techniques developed originally watershed delineation from DEMs. The approach worked very well and will be useful to structural engineers and architects who are studying masonry walls and want to quickly generate a finite element model of existing structures.
  • An evaluation of data from a multi-spectral lidar system (Optech Titan) was presented to show the capabilities and limitations. A novel human portable scanning solution without requiring an IMU was also presented.
  • Additional algorithms for point cloud de-noising and full-waveform processing were also unveiled.

Hope to see you at LaserScanning 2017!

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.