lidar calibration and validation software and processes

Ayman F. Habib, 2010

LiDAR Calibration and Validation Software and Processes

http://dprg.geomatics.ucalgary.caDepartment of Geomatics Engineering

University of Calgary, Canada1

http://dprg.geomatics.ucalgary.ca/


Acknowledgement

McElhanney Consulting McElhanney Consulting Services Ltd., Vancouver, BC, Services Ltd., Vancouver, BC,

CanadaCanada


Overview

• LiDAR systems: QA/QC• LiDAR system calibration

– Simplified calibration procedure– Quasi-rigorous calibration procedure– Evaluation (QC) criteria

• Relative accuracy• Absolute accuracy

– Experimental results• Concluding remarks


Quality Assurance & Quality Control

• Quality assurance (Pre-mission): – Management activities to ensure that a process, item, or

service is of the quality needed by the user– It deals with creating management controls that cover

planning, implementation, and review of data collection activities.

– Key activity in the quality assurance is the calibration calibration procedureprocedure.

• Quality control (Post-mission):– Provide routines and consistent checks to ensure data

integrity, correctness, and completeness– Check whether the desired quality has been achieved


• QA activities/measures include:– Optimum mission time– Distance to GNSS base station – Flying height– Pulse repetition rate– Beam divergence angle– Scan angle– Percentage of overlap– System calibration

LiDAR QA


00

lulb

blu

mbG

mboG RRRPRXX

6

System Calibration

LiDAR QA: System Calibration



• The calibration of a LiDAR system aims at the estimation of systematic errors in the system parameters.– One can assume that the derived point cloud after

system calibration is only contaminated by random errors.

• Usually accomplished in several steps:– Laboratory calibration,– Platform calibration, and– In-flight calibration


• Drawbacks of current in-flight calibration methods:– Some techniques require the raw data, which is not always

available.– Time consuming and expensive– Generally based on complicated and sequential calibration

procedures– Require some effort in ground surveying of the control

points/surfaces– Some of these calibration procedures involve manual and

empirical procedures.– Lack of a commonly accepted methodology



Error sources Error sources analysis / analysis /

Error Error ModelingModeling

Recoverability Recoverability analysisanalysis

Flight Flight configurationconfiguration

Sampling Sampling DensityDensity

CorrespondenceCorrespondence

PrimitivesPrimitives

Aspects Aspects InvolvedInvolved



• Calibration/system parameters:– Spatial and rotational offsets between various system

components (ΔX, ΔY, ΔZ, Δ, Δ, Δ)– Range bias (Δ)– Mirror angle scale (S)

• The system parameters can be estimated using the original LiDAR equation (rigorous approach).– Raw measurements should be available.

• These parameters can be estimated using a simplified version of the LiDAR equation (approximate approach).– Raw measurements need not be available.



• Quality control is a post-mission procedure to ensure/verify the quality of collected data.

• Quality control procedures can be divided into two main categories:– External/absolute QC measures: the LiDAR point cloud

is compared with an independently collected surface.• Check point analysis

– Internal/relative QC measures: the LiDAR point cloud from different flight lines is compared with each other to ensure data coherence, integrity, and correctness.

LiDAR QC


• LiDAR data is usually acquired from parallel flight lines with some overlap between the collected data.

• DPRG Concept: Evaluate the degree of consistency among the LiDAR footprints in overlapping strips.

Strip 2 Strip 3 Strip 4

LiDAR QA/QC: DPRG Approach


•LiDAR Data in Overlapping Parallel Strips

Point cloud coordinates Raw measurements are not necessarily

available

LiDAR QA/QC: DPRG ApproachSimplified Calibration


Overlapping strips

Discrepancies

3D Transformation

Rotation

ShiftsCalibration Parameters

•LiDAR Data in Overlapping Parallel Strips

Point cloud coordinates Raw measurements are not necessarily

available

QC ProcedureQA Procedure



YM

ZM

XM

Zlu

Ylu

Xlu

Zlu

Ylu

Xlu

Mapping Frame

Laser Unit Frame

Laser Unit Frame

PZ

Y

X

User Defined Coordinate

System

15


Local coordinate system


},,,,,,{},,,{ SYXrollZYX TTT

Overlapping strips

Discrepancies

Rigid body Transformation: Three translations and a roll angle



Dx =0

Zlu

Ylu

Xlu

Zlu

Ylu

Xlu

Laser Unit Frame

Laser Unit Frame

Strip A

Strip B

Z

Y

X


System

Backward

Forward

xA =-x

B

P

'2

'

02222

),,,(

BG

AG XRHY

HXX

andYXf


17



Dx

Zlu

Ylu

Xlu

xA

xB

Laser Unit Frame

PStrip AStrip B

Z

Y

X


System

Forward

Zlu

Ylu

Xlu

Laser Unit Frame

Forward

'2

')/(

),,,(

BGS

HD

X

X

XXAG XR

DD

SDHDX

andSf

X

18


LiDAR QA/QC: DPRG ApproachQuasi-Rigorous Calibration

19


•LiDAR Data in Overlapping Strips Point cloud coordinates with the time tag Time-tagged trajectory


Trajectory point Estimated

trajectory point

Fitted trajectory line

Time-tagged laser pointx(t)

z(t)

N

κ(t)

β(t)

P( t)

ti ti+1

ti+2

ti+3

20


AB

BBAA

BBBAAA

BBBAAA

BA

BBAA

BBAA

BBAA

BBAA

BA

BBAABBAA

BBAABBAA

BABA

BABA

BiasedB

BiasedA

BiasedB

BiasedA

BiasedB

BiasedA

eSxx

SzzSzz

SSSSSS

xxxx

xxzzzz

zzzz

YXYX

ZZYYXX

)(sinsincoscos

coscossinsinsinsinsincossincos

0coscossinsin

)(sinsincoscos

coscossinsin

0coscossinsinsinsincoscos

~~~~~~

Assuming that A and B are conjugate points



AB

BB

BBB

BBB

B

BB

BB

BB

BB

B

BBBB

BBBB

BB

BB

BiasedB

ControlA

BiasedB

ControlA

BiasedB

ControlA

eSx

SzSz

SSS

xx

xzzzz

ZYXYX

ZZYYXX

sin cos

cos sinsin

sincos

0 cos

sin

sin cos

cos sin

cos sin sin cos

~~~~~~

Assuming that A and B are conjugate points



Optimum Flight Configuration



Conjugate patch to a given point

Point/Patch Pairs: Closest Patch Procedure

Conditions:• Closest patch (within

a threshold)• Point located within

the patch


• Procedures have been developed to deal with the absence of corresponding points within conjugate point-patch pairs.


• Relative Accuracy– Qualitative Evaluation:

• Intensity images before and after the point cloud adjustment• Profiles before and after the point cloud adjustment• Segmented point cloud

– Quantitative Evaluation:• Average noise level within segmented point cloud• Discrepancies between overlapping strips before and after the point

cloud adjustment

• Absolute Accuracy– LiDAR features, derived from the original and adjusted point

cloud, are used for photogrammetric geo-referencing– Check point analysis

25

Evaluation Criteria


Strip Number

Flying Height Direction

1 1150 m N-S

2 1150 m S-N

3 539 m E-W

4 539 m W-E

5 539 m E-W

6 539 m E-W

Strip 1

Strip 2

Strip 3

Strip 4

Strip 5

Strip 6

Strip 1

Strip 2

Strip 3

Strip 4

Strip 5

Strip 6

Experimental Results

Data Captured by ALS50


OverlappingStrip Pairs Overlap % Direction

Strips 1&2 80% Opposite directions



Strips 5&6 50% Same direction




Estimated biases in the system parameters



Original Point Cloud

1m1m

Impact on Generated Profiles



Adjusted Point Cloud

1m1m

Impact on Generated Profiles


Before Calibration After Calibration

Strips 1&2 Strips 1&2

XT (m) YT (m) ZT (m) XT (m) YT (m) ZT (m)

1.10 -0.32 -0.01 0.10 0.07 -0.05

(deg) φ (deg) κ (deg) (deg) φ (deg) κ (deg)

0.0001 -0.052 -0.002 0.0012 -0.0016 -0.0055

Strips 3&4 Strips 3&4

XT (m) YT (m) ZT (m) XT (m) YT (m) ZT (m)

0.18 0.41 -0.01 -0.01 0.03 0.00

(deg) φ (deg) κ (deg) (deg) φ (deg) κ (deg)

0.0484 -0.0005 -0.0011 0.0075 0.0009 -0.0013

Compatibility between overlapping strips before and after the calibration procedure

31

Experimental ResultsImpact on Existing Discrepancies


Photogrammetric Data

Camera model Rollei P-65

Array dimension 8984x6732 pixels

Pixel size 6μm

Nominal focal length 60mm

Camera classification Normal Angle Camera (AFOV = 58.6º)

Camera Specifications

• Six flight lines:• Four parallel flight lines @ 550m (50%

side lap)• Two opposite flight lines @ 1200m

(100% side lap)

L1 L1

L2 L2

L3L3

L4L4

L5L6

L5L6

Impact on Absolute AccuracyExperimental Results


Before Calibration After Calibration

Mean ΔX (m) -0.03 0.02

Mean ΔY (m) -0.18 0.01

Mean ΔZ (m) 0.15 0.08

σX (m) 0.11 0.05

σY (m) 0.15 0.06

σZ (m) 0.17 0.18

RMSEX (m) 0.11 0.06

RMSEY (m) 0.23 0.06

RMSEZ (m) 0.23 0.19

RMSETOTAL (m) 0.34 0.21

RMSE analysis of the photogrammetric check points using extracted control linear features from the LiDAR data before and after the

calibration procedure

Experimental ResultsImpact on Absolute Accuracy


Concluding Remarks• In spite of the technical advances in LiDAR

technology, there is still a lack of well defined procedures for the Quality Assurance (QA) and Quality Control (QC) of the Mapping process.

• These procedures should be capable of the dealing with the nature/restrictions of the current mapping procedure.– Absence of the system raw measurements– Challenge in having LiDAR specific control targets

• This research has developed a calibration procedure that led to improvements in the relative and absolute accuracy of the adjusted point cloud.


Comments and Questions?

lidar calibration and validation software and processes

Documents