Recent developments in laser scanning Kourosh Khoshelham With contributions from: Sander Oude Elberink, Guorui Li, Xinwei Fang, Sudan Xu and Lucia Diaz Vilarino
Why laser scanning? Laser scanning accurate capturing of 3D geometry Point cloud; Point clouds suitable for automated processing; Automated processing = Fast, inexpensive, less labor intensive. Terrestrial Laser Scanning (TLS) Mobile Laser Scanning (MLS) Aerial Laser Scanning (ALS) 2
Overview of developments in laser scanning (by no means exhaustive) Platforms Developments Hardware technology ALS TLS MLS Higher pulse/scan frequency Multiple pulses in the air (MPiA) Full-waveform recording UAV-based laser scanning Smaller, more userfriendly scanners Combined phase/pulse range measurement Full waveform recording SLAM systems (mobile mapping in GPS-denied environments) Processing methodology Intensity calibration Waveform processing Data integration and fusion with other sources Automated feature/object extraction Cloud computing Waveform processing Data integration and fusion with other sources Automated feature/object extraction Cloud computing Data integration and fusion with other sources Automated feature/object extraction Cloud computing Applications 3D change detection Damage mapping Coastal monitoring Water management Indoor mapping Non-topographic applications: Geology, Hydrology, Road furniture inventory Railway asset management Indoor mapping 3
Developments in hardware technology: Multiple Pulses in the Air (MPiA) Original one pulse in the air: Reduces pulse rate at high altitudes to avoid range ambiguity; Multiple pulses in the air (MPiA) technology: Allows higher pulse rates at high altitudes From Leica ALS60 Product Specifications 4
Developments in hardware technology: Multiple Pulses in the Air (MPiA) Year of introduction 2011 2010 2012 2010 Multiple pulses in air Yes Yes Yes Yes Data from: http://www.geo-matching.com/ 5
Full-waveform recording ALS Year of introduction 2011 2010 2012 2010 Full-waveform digitization Yes Yes Yes Yes TLS Year of introduction 2010 2010 2010 2010 Full-waveform digitization No Yes No No 6
More versatile terrestrial laser scanners Smaller and lighter scanners; Better accuracy at longer range using wavepulse technology (hypermodulation); More user-friendly interface (smartphone); Camera option. Year of introduction 2010 2010 Total weight (kg) 11.8 5.0 Ranging principle Phase+Pulse Phase (hypermodulation) Max range (m) 80 120 Camera option Yes Yes User interface Tablet PC, Tablet, Smartphone 7
Developments in SLAM Mobile mapping in GPS-denied (indoor) environments; IMU errors corrected by other means (odometer, scene structure, loop closure). TMMS Zebedee From: http://www.lidarnews.com/ 8
Honorable mention: Kinect RGB-D cameras like the Kinect have great potential for indoor mapping; Kinect captures: depth + color images @ ~30 fps = sequence of colored point clouds But: lower accuracy and lower depth resolution compared to laser scanning IR emitter RGB camera IR camera + 9
Developments in processing methodology and applications: Rail track detection and modelling 10
Rail track detection and modelling Detection based on rail properties: Slightly above ground; Contact wires above at a certain height; Locally linear; Modeling by: Fitting small rail pieces; Parameter estimation by MCMC (robust to gaps and outliers); Interpolation by smooth Fourier curves. 11
Developments in processing methodology and applications: Rail track detection and modelling 12
Rail track detection and modelling Point-model distances: Accuracy ~2 cm, good enough for visualisation, asset inventories; Not for detection of rail wear 13
Road furniture inventory Supervised classification of connected components Accuracy: ~ 85% Work of Guorui Li Tree Lamp post Car Facade Other 14
Road furniture inventory Detection of moving objects by comparing two sensor datasets Accuracy: ~ 90% Sensor 1 Sensor 2 Work of Xinwei Fang 15
3D change detection Comparing point clouds from 2008, 2010 and 2012; Based on analyzing distances between closest points. Work of Sudan Xu 16
Damage mapping 17
Smart cities: sensor networks coupled with 3D models 3D city model Indoor model Image from ESRI City Engine 1 st floor 2 nd floor Model from Google 3D warehouse 18
(Semi-) automated modelling of indoor environments Oriented point cloud Work of Lucia Diaz Vilarino 19
(Semi-) automated modelling of indoor environments Segmented point cloud Work of Lucia Diaz Vilarino 20
(Semi-) automated modelling of indoor environments Intersecting adjacent planes vertices Work of Lucia Diaz Vilarino 21
(Semi-) automated modelling of indoor environments Model with faces and vertices 22
(Semi-) automated modelling of indoor environments Textured model 23
Summary, future trends and challenges Laser scanning already an established technology; But still developing at a steady pace: Laser scanner hardware technology (MPiA, full-waveform, ); Processing methodology (automated information extraction); Applications (roads, railroads, indoor environments, ). Future trends: Indoor mapping systems, SLAM; Big Data, not only handling but learning from it more automation in information extraction; Cloud computing;? Challenges in practice:? 24