REMOTE SENSING LiDAR & PHOTOGRAMMETRY19 May 2017

SERVICES

Visual Inspections

Photo maps

Digital Terrain Models

Thermal Inspections

Aerial Imagery Volume Computations

Aerial Video Training & Consultancy

SYSTEMS

3

MD4-1000

HEF-30 (2x)

Zenith (2 x)

DJI Inspire I (2x)

Asctec Falcon V8 (3x)

Trimble UX 5 HP

Cessna Balloon (5x)

SkeyeBat

CLIENTS

UAV LiDAR vs PHOTOGRAMMETRY

5

🥊🥊

LiDAR PRINCIPLE

6

Distance = Time of travel / 2

Speed of light

Transmitter

Receiver Reflector

BATHYMETRIC LiDAR

7

LiDAR PRINCIPLE ACTIVE LIGHT

8

POSITIONING LIDAR

9

10

POSITIONING LIDAR

11

POSITIONING LIDAR

12

POSITIONING LIDAR

13

POSITIONING LIDAR

POSITION AND ORIENTATION ERRORS

ARE NOT THE SAME FOR ALL

RETURNS

PER SCAN

=> NOT CORRELATED

LiDAR ERROR SOURCES

14

Sensor Position

GPS error

INS/IMU error

GPS-IMU Integration error

Angular Errors

Misalignment between LiDAR scanner and IMU (Boresight

calibration)

Lever arm Error

Incorrect positioning between GPS antenna and LiDAR sensor

LiDAR Range Error

Precision of LiDAR scanner

Divergence of Laser beam

Multipath error

Reflection on a sloping surface

LiDAR ERROR SOURCES

15

Range

Between 5 mm to 20 mm

Position

With RTK or PPP Positioning between 15 mm and 50 mm

Orientation

Between 0.025 degrees and 0.15 degrees

Example Sum of all errors

Velodyne HDL 32E Scanner

Flying Height 60 meters AGL (Above Ground Level)

Range error: <= 20 mm

GNSS Positioning

Horizontal: 1 cm + 1ppm, assume 11mm

Vertical : 1.5 times horizontal = 16.5 mm

Total = √(11 mm2 + 16.5 mm2) = 19.83 mm

Range and Positioning error: 20 mm + 19.83 mm = 39.83 mm

IMU accuracy Pitch and roll: 0.15° ⟹ 60 meters Range = 60 * tan(0.15°) = 15.7 cm

IMU accuracy Pitch and roll: 0.025° ⟹ 60 meters Range = 60 * tan(0.015°) = 2.62 cm

Total Error = √(15.72 + 3.92) = 16.18 cm / Total Error = √(2.622 + 3.92) = 4.7 cm

LiDAR PROJECT SCHEVENINGEN BREAKWATER

PHOTOGRAMMETRY

17

PHOTOGRAMMETRY

18

PHOTOGRAMMETRY

19

PHOTOGRAMMETRY

20

POSITION AND ORIENTATION ERRORS

ARE THE SAME FOR ALL PIXELS

PER PHOTOGRAPH

CORRELATED

NOT WITH ROLLING SHUTTER !!

Rolling Shutter and Photogrammetry

21

STEREO VIEWING

22

PHOTOGRAMMETRY

PHOTOGRAMMETRY

PHOTOGRAMMETRY ALLIGNMENT

PHOTOGRAMMETRY

PHOTOGRAMMETRY

PHOTOGRAMMETRY ACCURACIES

General ‘rules of thumb’ for photogrammetry with dense matching techniques

- Relative accuracy is influenced by resolution (GSD, Ground Sampling Distance)

- Absolute accuracy is influenced by quality of the geodetic network (i.e. ground control points)

- Absolute accuracy is influenced by the data processing methodology

- If all of the above are favorable:

- X,Y accuracy is 1 to 1.5 times the GSD

- Z accuracy is 1.5 to 2 times the GSD

- Absolute accuracy is the quality of the network + relative accuracy

Sample project Scheveningen breakwater

- Flight altitude 40 meters with Sony A7r (36 Mp and 35mm lens) => GSD = 0.7 cm

- Quality of the Ground control points assumed at 2cm X,Y and 3 cm Z

- A priori estimated error = √((1.5 ∗ 0.7)2 + 32) = 3.18 cm

PHOTOGRAMMETRY SAMPLE PROJECT

PHOTOGRAMMETRY SAMPLE PROJECT

PHOTOGRAMMETRY SAMPLE PROJECT

PHOTOGRAMMETRY ACCURACIES BREAKWATER SCHEVENINGEN

X Y

Height

Level GPS DEM Dz-1 Dz-2 Absolute Dz-1 Absolute Dz-2

GCP01 77542.555 457425.012 5.676 5.681 5.686 0.005 0.010 0.005 0.010

GCP02 77519.250 457437.892 5.117 5.117 5.118 0.000 0.001 0.000 0.001

GCP03 77524.464 457471.887 4.607 4.623 4.62 0.016 0.013 0.016 0.013

GCP04 77534.839 457515.828 5.557 5.564 5.564 0.007 0.007 0.007 0.007

GCP07 77482.622 457470.247 4.542 4.544 4.541 0.002 -0.001 0.002 0.001

GCP08 77455.233 457499.366 4.525 4.533 4.523 0.008 -0.002 0.008 0.002

GCP14 77326.597 457699.824 4.519 4.521 4.516 0.002 -0.003 0.002 0.003

GCP20 77285.905 457852.778 4.511 4.509 4.511 -0.002 0.000 0.002 0.000

GCP23 77283.584 457876.050 4.496 4.502 4.494 0.006 -0.002 0.006 0.002

Average 0.005 0.003 0.005 0.004

STDEV 0.005 0.006 0.005 0.005

Dz-1 = Difference Level - GPS

Dz-2 = Difference Level - DEM

LiDAR vs PHOTOGRAMMETRY (UAV ONLY!)

LiDAR

✔ Vegetation Penetration

✔ Detect smaller features (i.e. power line)

✔Quicker data processing

✔ No (or little) Ground control

✔ Active light (better in dark/shadow areas)

✖ No Picture

✖ Accuracy

✖ Cost

✖ Weight (i.e. safety)

Photogrammetry

✔ Accuracy

✔ Costs

✔Weight

✔ Picture

✖ Only map what you see

✖ Longer Processing times

✖ Cannot detect small features

✖ Ground Control (even with RTK or PPK!)

✖ Less accurate in shadow areas

CONCLUSION:

One sensor is not ’better’ than the other. Depends very much on the type of project.