webinar on introduction to 3d gis data gepsus set• vermessung avt gmbh – founded in 1970 –...

64
www.terra-messflug.at Webinar on Introduction to 3D GIS Data – GEPSUS set 2-5 December 2014 Trento, Imst, Podgorica

Upload: others

Post on 25-Jan-2021

2 views

Category:

Documents


0 download

TRANSCRIPT

  • www.terra-messflug.at

    Webinar on Introduction to 3D GIS Data – GEPSUS set

    2-5 December 2014

    Trento, Imst, Podgorica

  • • Presentation of Terra Messflug and AVT

    • Introduction to aerial photogrammetry

    • Workflow for the production of Trento dataset

    • Webinar organization

    • Discussion

    Outline

    02.12.2014 GEPSUS Introduction 2

  • • Vermessung AVT GmbH – Founded in 1970

    – Head-office in Imst, Austria, 65 employees

    – Activities: cadaster, topography, geodesy, photogrammetry, aerial and terrestrial laser scanning, geoinformation

    Vermessung AVT

    02.12.2014 GEPSUS Introduction 3

  • • Terra Messflug – 100% subsidiary of Vermessung AVT

    – Focused in aerial digital image acquisition and processing in Central Europe

    • Aircraft: Cessna T303

    • Optical sensor: UltraCam Xp

    • GSD da 3 cm a 30 cm

    – Other services:

    • Oblique image acquisition

    • Lidar data acquisition (partners)

    • Lidar data processing

    • Thermal image acquisition

    • Thermal image processing

    Terra Messflug

    02.12.2014 GEPSUS Introduction 4

  • • Photogrammetry is the science of making measurements from photographs

    – 'photo': light

    – 'gram‘: drawing

    – 'metry‘: measurement

    • Measurements: coordinates, distances, areas, volumes

    • Photogrammetric products:

    – Digital Terrain / Surface models

    – Orthoimages

    – 2D / 3D Vectors

    – Other GIS data

    What is photogrammetry

    02.12.2014 GEPSUS Introduction 5

  • • According to the platforms, we distinguish: – Spaceborne photogrammetry: the

    camera is mounted on a spacecraft

    – Aerial photogrammetry: the camera is mounted on an airplanes, helicopter of UAV

    – Close-range (terrestrial) photogrammetry: the camera is close to the object (hand, tripod)

    What is photogrammetry

    02.12.2014 GEPSUS Introduction 6

  • Optical sensors on airborne platforms

    02.12.2014 GEPSUS Introduction 7

  • • Analogue cameras: – Images are stored on films

    – Films are developed

    – Images are printed on paper or dia

    – Image are processed using digital/analytical image stations

    • Digital cameras: – Images are acquired using CCD or CMOS sensors

    – Images are stored on a digital device

    – Images are processing using digital photogrammetric stations

    Analogue / Digital

    02.12.2014 GEPSUS Introduction 8

    PA

    ST

    T

    OD

    AY

  • • The optical sensor consists of: – Optical system (lenses)

    – One or more CCD or CMOS matrices (frames) in the focal plane

    – In case of more matrices, the single images are assembled

    – Multiple bands for multispectral acquisition

    Frame cameras

    02.12.2014 GEPSUS Introduction 9

  • • For the extraction of 3D information (planimetric and height), stereo photogrammetry is used

    • Stereo images: images of the same object acquired from different directions -> cameras are separate by a base B

    Stereo photogrammetry

    02.12.2014 GEPSUS Introduction 10

    O1: position of camera 1 O2: position of camera 2 P´: homologous point P in image 1 P´´: homologous point P in image 2 B: base of the stereopair

  • • In aerial photogrammetry: in blocks – Min along-flight overlap between images: 60%

    – Min side overlap between strips: 30%

    How are stereo images acquired?

    02.12.2014 GEPSUS Introduction 11

    strip

    strip

    strip

    strip

  • Orientation = „Reconstruction of the rays from the object point P to the image point P´, P´´, etc. ”

    In general, to extract 3D information from images, we need to know the position and viewing direction of the camera during image acquisition.

    Orientation: Interior O. and Exterior O.

    with:

    Interior Orientation = Reconstruction of the path of the ray inside the

    acquisition device = camera geometry

    Exterior Orientation = Reconstruction of the position during image

    acquisition = camera position and attitude

    What is sensor orientation

    02.12.2014 GEPSUS Introduction 12

  • 1. Describe the relationship between image and ground coordinates of the same point through equations that contain image and ground point coordinates, interior and exterior orientation parameters -> collinearity equations

    2. Write collinearity equations for a certain number of points with known ground coordinates (ground control points - GCPs) and other tie points (TPs) with unknown ground coordinates

    3. Build a system of equations (2 collinearity equations for each point). The unknowns of the system are parameters describing the exterior and interior orientation of the camera

    4. Solve the system with least-square adjustment using GCPs and TPs. The images are oriented.

    5. Once the images are oriented, we can calculate the 3D coordinates of any homologous point in the images

    How do we orient images

    02.12.2014 GEPSUS Introduction 13

  • Assumption:

    - Images are the result of a central projection

    - Images are generated on the focal plane of the camera

    Geometric model and interior orientation

    02.12.2014 GEPSUS Introduction 14

    A

    D

    B

    C

    A’

    D’

    B’

    C’

    LENS (projection

    centre)

    OBJECT SPACE

    Focal place

    IMAGE SPACE

    Focal length

    Principal Point PP

  • )()()(

    )()()(

    )()()(

    )()(`

    C33C23C13

    C32C22C120

    C23C23C13

    C31C21C110

    ZZrYYrXXr

    ZZrYYrXXrfyy

    ZZrYYrXXr

    ZZrYYrXXrfxx

    R(C,C,C): rotation matrix from image space to object space containing orientation angles

    CCC ZYX : camera position in object space

    y]x[ : image coordinates

    with: X

    Y

    Z x

    y z

    P

    PC(XC, YC, ZC)

    P‘

    C

    C C

    x0, y0 : principal point; f : focal length

    f

    y

    x

    kR

    Z

    Y

    X

    Z

    Y

    X

    CCC

    C

    C

    C

    ),,(

    • Relation between image and ground coordinates

    External orientation

    02.12.2014 GEPSUS Introduction 15

    Collinearity equations

    x0, y0

    Ground reference system

  • Focal plane

    Lens or mirror system

    Image acquisition device,

    e.g. frame- or line- sensor

    Object, e.g. earth surface

    c = Focal distance

    Source of error:

    -Imperfectness during the manufacturing process of

    the device and the lens system, e.g. imperfect

    grinding and decentration of the lenses from the

    optical axis

    -Unevenness of the focal plane

    Additional sources of error for digital sensors:

    -Pixel spacing and positioning

    -Problems during the read-out process

    -Noise

    -Blooming

    Problem: Image acquisition using a camera or other devices is not a perfect projection

    => The ray from the object space to the image space is not straight. ε1 ε2

    ε3

    Solution: Modeling and consideration of the errors:

    ε1 = Lens distortions

    ε2 = Shift of the principle point

    ε3 = Error in the focal length

    Interior orientation

    02.12.2014 GEPSUS Introduction 16

  • - Position of principal point(s)

    - Change of focal length

    - Radial distortion

    - Decentering

    - Shear

    - Scale

    • Additional parameters for the estimation of

    x21222412P sxaypyx2px2rxkrkrxf

    fxx

    yP syaxpyrpyxykrkryf

    fyy

    2

    22

    12

    4

    1

    2 22

    pp xxx

    pp yyy

    2p

    2p

    2 yxr

    Interior orientation

    02.12.2014 GEPSUS Introduction 17

  • y)ZZ(r)YY(r)XX(r

    )ZZ(r)YY(r)XX(rfyy

    x)ZZ(r)YY(r)XX(r

    )ZZ(rYYr)XX(rfxx

    C33C23C13

    C32C22C120

    C23C23C13

    C31C21C11`0

    • Extended collinearity equations

    pp xxx

    pp yyy

    2p

    2p

    2 yxr

    Interior orientation

    02.12.2014 GEPSUS Introduction 18

    x21222412P sxaypyx2px2rxkrkrxf

    fxx

    yP syaxpyrpyxykrkryf

    fyy

    2

    22

    12

    4

    1

    2 22

    with

  • • Laboratory calibration aims at estimated the interior orientation parameters of the camera.

    • The calibration target includes a number of signalized points with known position (GCPs), distributed in the camera field of view at different depth.

    Camera calibration

    02.12.2014 GEPSUS Introduction 19

    • Images are taken and the parameters estimated in a bundle adjustment

    • A set of parameters is estimated for each lens

    C:/Users/Daniela Poli/Documents/DanielaPoli/Sales/Trento_Graphithec_2014/GEPSUS/Bid_2014_03/Webinar/CameraCalibration_UCxp_V80_ShortVersion.pdf

  • • During image acquisition, geopositioning and inertial systems are mounted and synchronized with the camera. In case of Terra: – Double frequency GPS NovAtel OEM I/IV

    – Inertial system IMU-Ild from IGI (Ingenieur-Gesellschaft für Interfaces mbH, http://www.igi.eu/ ), Germany

    • GPS / IMU measure directly the trajectory of the airplane

    Approximation for exterior orientation

    02.12.2014 GEPSUS Introduction 20

    • After the flight, the trajectories are processed with differential technique (DGPS) with master stations (max distance from images 50km) and the approximate position and attitude of each image are estimated.

    http://www.igi.eu/http://www.igi.eu/

  • • The geometric calibration of IMU system is executed through these steps: – Visual test for damages of the case and connectors

    – Verification of the faultless function of each subsensor

    – Determination of special coefficients for trajectory modelling.

    • In case of Terra, the calibration is executed by IGI and coefficients for data processing with AEROoffice are provided

    • Statistics on the accuracy of the position and attitude estimation are given.

    IMU calibration

    02.12.2014 GEPSUS Introduction 21

    C:/Users/Daniela Poli/Documents/DanielaPoli/Sales/Trento_Graphithec_2014/GEPSUS/Bid_2014_03/Webinar/IGI_IMUKalib_2013-12-11.pdf

  • Workflow for Trento dataset

    Pre-flight

    • Flight planning

    • Permission requests

    • Camera calibration

    • GPS/IMU calibration

    Image acquisition

    • Aerial flight

    • Data storage

    Image preparation

    • Radiometric processing

    • Trajectory estimation

    Image orientation

    • GCPs measurement

    • Aerial triangulation

    3D Modeling

    • Dense image matching

    • 3D point cloud generation

    • Classification

    • Generation of DSM and DTM

    Orthophoto production

    • True-orthophoto

  • Flight planning – operational area

    02.12.2014 GEPSUS Introduction 23

    Home airport:

    • Zell am See (Austria)

    • Aalen (Germany)

    Operating distance:

    • 1.500 km / 2.000 km

    Imst

  • Aircraft Cessna C303

    02.12.2014 GEPSUS Introduction 24

    Characteristic

    values:

    • Twin-engine

    • 125 – 400 km/h

    • 7.600 m max.

    flying altitude

    • 6.5 hours max.

    endurance

  • Chief-Pilot: Marko Schnelzer • Aircraft owner

    • Pilot and camera operator for > 20 years

    • > 7.000 operational hours of photo flight missions

    • One of the most experienced pilots in alpine environments

    Pilot: Mario Königstorfer • Pilot for 15 years

    • > 2.700 operational hours of photo flight missions

    License: Air Worthiness Certificate – CAMO

    In Italy: license from ENAC and ENAV national authorities

    Pilot’s experience and license

    02.12.2014 GEPSUS Introduction 25

  • • Image acquisition is planned in order to fulfill the project requirements: – Ground sample distance: 10cm

    – Overlap: 80% along, 60% side. This is required for dense image matching and the generation of a dense 3D point cloud

    • A rough digital terrain model is used to calculate the image footprints on the terrain

    Flight plan

    02.12.2014 GEPSUS Introduction 26

  • • Effect of DTM during flight planning

    Flight plan

    02.12.2014 GEPSUS Introduction 27

    Rovereto, 2013, GSD:5cm, overlap 80%, 60%

    without terrain model

    (flat terrain supposed) with terrain model

  • Flight plan Trento

    02.12.2014 GEPSUS Introduction 28 Coverage of flight lines

  • Camera: • UltraCam Xp (digital frame camera)

    Data-units: • 3 data-units (fully redundant storage on 2 data-units)

    • Storage capacity for 6.500 images each

    INS / IMU: IGI AeroControl IMU II-d 256 Hz

    GPS: NovAtel OEM V (multi-frequency-receiver)

    VRF, IFR, Radar

    CCNS4

    Telescope and T-AS

    Equipment in aircraft

    02.12.2014 GEPSUS Introduction 29

  • Camera: UltraCam XP

    02.12.2014 GEPSUS Introduction 30

    Specifications:

    • 17.310 x 11.310 pixels

    (196Mp)

    • max. GSD: 2.8 cm

    • Focal length: 100.5 mm

    • Geometric accuracy: 2 µm

    • Channels: PAN, RGB, NIR

    • Colour depth: >12 bit

    • Colour depth during

    processing: 8 or 16 bit

    • Regular calibration and

    maintenance

  • Photo flight mission

    02.12.2014 GEPSUS Introduction 31

  • • Consideration of all relevant air-traffic regulations during flight planning and during flight execution

    Flight management: Air-traffic control

    02.12.2014 GEPSUS Introduction 32

  • • Good weather conditions are prerequisite for good images

    • Up to hourly checks of weather conditions (forecast and real-time)

    • allows for efficient medium-term flight mission planning

    • allows for quick reaction in changing weather conditions

    Flight management: weather conditions

    02.12.2014 GEPSUS Introduction 33 Weather forecast: e.g. Deutscher Wetterdienst

    Real-time weather check using webcams

  • Trento flight in brief

    02.12.2014 GEPSUS Introduction 34

    Digital camera Vexcel Ultracam-Xp Average GSD 10 cm Along-track overlap 80% Side overlap 60% Number of strips 9 Number of images 397 Spectral channels Red, Green, Blue, Near Infrared Radiometric resolution 12 bit Date of acquisition 23 September 2013 Area covered by the flight 55 km² Urban area covered by the flight 18 km²

  • • Manpower, hard- and software

    • 16 experts involved in flight management, image processing, photogrammetry and subsequent analysis

    • High performance fiber-channel IBM SAN storage (ca. 100 T) • High performance computation cluster • ca. 25 IBM Quad-/dualcore high-performance PCs • Calibrated and 3D monitors

    • Topoflight • Microsoft UltraMap • Waypoint GrafNav, IGI AeroOffice • Inpho Geo-Imaging (3 licenses) • OrthoMaster, Orthovista, Orthovista Seam Edit, Applications Master • Match-T, SCOP++ • Global Mapper, ArcGIS • DAT/EM Summit Evolution (2 licenses) • CAD: Bentley Microstation (4 licenses), Autodesk AutoCad Map (6 licenses)

    Overview infrastructure Imst

    02.12.2014 GEPSUS Introduction 35

  • Hardware:

    • High-performance IBM SAN storage system (100 TB)

    • High-performance computer cluster for image processing (up to 32 CPU-cores in parallel)

    • Transportable data storage devices (12 TB)

    Data storage and processing

    02.12.2014 GEPSUS Introduction 36

  • • Differential correction of GNSS positioning using permanet receiving GNSS stations and processing of INS

    Computation of flight trajectory

    02.12.2014 GEPSUS Introduction 37

    Utilised software:

    • AeroOffice

    • GrafNav

  • Software:

    • Microsoft UltraMap

    • DragonFly image format: visualisation of huge image blocks (> 5000 images)

    • project-based-colour-balancing to achieve homogeneous radiometric setting for the entire project area

    Radiometric processing

    02.12.2014 GEPSUS Introduction 38

  • • Manual radiometric adjustments are done for each image to optimise image quality

    • Utilised software: UltraMap Radiometry

    Manual radiometric adjustment

    02.12.2014 GEPSUS Introduction 39

  • PBCB substantially improves image block homogeneity

    Project Based Colour Balancing (PBCB)

    02.12.2014 GEPSUS Introduction 40

    before PBCB after PBCB

  • • Each project is evaluated regarding radiometric homogeneity of the entire project area

    Quality control in geometrical context

    02.12.2014 GEPSUS Introduction 41

    Image block consisting of >140 images

    • GSD = 10 cm

    • Overlap = 60%, 30%

  • • The purpose of aerial triangulation is to improve the trajectory measured by the instruments onboard and estimate the position and attitude of each image.

    • Ground control points were surveyed with centimeter precision and used as input for aerial triangulation

    • Software used: Match-T by Inpho

    Aerial triangulation

    02.12.2014 GEPSUS Introduction 42

  • • Software suite: Inpho Geo-Imaging 5.6

    Aerial triangulation

    02.12.2014 GEPSUS Introduction 43

    Licences and

    components

    • 3 licences

    • OrthoMaster

    • Orthovista

    • Orthovista Seamedit

  • • Ground control points for absolute orientation of images

    Aerial triangulation

    02.12.2014 GEPSUS Introduction 44

    Procedure

    1. Planning in office

    2. Field measurements

    3. Computation

  • • Quality control

    Aerial triangulation

    02.12.2014 GEPSUS Introduction 45

    Procedure

    • Statistics from software

    • GCPs

  • • Image matching: establishment of consistent correspondences between primitives extracted in two or more images, and estimation of the corresponding 3D coordinates, using collinearly models or projective models.

    • Many algorithms proposed, classified, for example, in – Stereo matching / multi-view matching

    – Area-based / Feature-based matching

    – Sparse / dense matching

    Image Matching

    02.12.2014 GEPSUS Introduction 46

  • • Innovative methodology for the generation of dense 3D point clouds

    • Point density up to pixel level – Trento dataset: up to 100 pt/m2 (matching every pixel)

    • Competitive to Lidar, in particular in urban areas

    • Software used: SURE by nFRAMES (http://nframes.com/ )

    • The 3D points are triangulated and interpolated in regular surfaces (DSM)

    Dense Image Matching (DIM)

    02.12.2014 GEPSUS Introduction 47

    http://nframes.com/

  • 3D point clouds

    02.12.2014 GEPSUS Introduction 48

  • DSM

    DTM

    Surface measurement and reconstruction:

    • DEM = Digital Elevation Model = any kind of elevation model

    • DSM = Digital Surface Model = surface with vegetation and man-made objects

    • DTM = Digital Terrain Model = surface without vegetation and man-made objects

    Digital surface models

    02.12.2014 GEPSUS Introduction 49

  • DIM in Trento

    02.12.2014 GEPSUS Introduction 50

  • DIM in Trento (with RGB information)

    02.12.2014 GEPSUS Introduction 51

  • • Ground / above-ground point classification

    • Generatin of digital surface model (DSM) and digital terrain model (DTM)

    DSM / DTM

    02.12.2014 GEPSUS Introduction 52

  • Aerial image:

    Not possible to measure correct

    distances, due to perspective effects

    Orthophoto:

    Measured distances, areas

    and positions are like on a

    map

    Aerial image:

    Overlapping of image and vector data

    is not consistent / correct

    Orthophoto:

    Correct overlap between image and vector

    data

    Orthophoto generation

    02.12.2014 GEPSUS Introduction 53

  • Aero triangulation

    • Highly precise matching and orientation of images (incorporating GPS/INS and ground GCPs)

    Digital elevation model

    • DTM for standard orthophotos

    • DSM for true-orthophotos

    Orthophoto

    • Geometric and radiometric corrected image block

    Orthophoto computation

    02.12.2014 GEPSUS Introduction 54

  • • Quality control

    Orthophoto

    02.12.2014 GEPSUS Introduction 55

    Problems in DEMs and necessary corrections: Manual seam line edit:

  • • Influence of sidelap between adjacent flights lines on image leaning

    Orthophoto

    02.12.2014 GEPSUS Introduction 56

    2009 - 60/30 %

    Example from orthophoto production in Denmark

    2013 - 80/60 % 2009 - 60/30 % 2013 - 80/60 %

  • • Images are projected on the DSM

    • In addition to terrain distortions, building distortions are corrected

    • Occluded areas between buildings are detected and filled by taking the missing image parts from neighboring images.

    • DSM-based orthophotos can be produced only if the aerial images have sufficient overlap in along and side direction. This is the case of Trento images.

    DSM-based orthophotos

    02.12.2014 GEPSUS Introduction 57

  • • Difference between…

    DSM/DTM-based orthophotos

    02.12.2014 GEPSUS Introduction 58

    …DTM-based orthophotos

    • Images rectified on terrain

    • Buildings facades and roofs appear

    displaced

    … and DSM-based orthophotos

    • Images rectified on terrain and objects on

    terrain

    • Building roofs appear on the footprints

    • Good for 3D visualization

  • Use of 2 (or more) oriented images and measure (MANUALLY or AUTOMATICALLY) identical

    points in the images for the determination of 3D object coordinates of the homologues points

    Known: orientation of (at least) two images

    Measured: image coordinates of a given point in both images (x1, y1, x2, y2)

    measurement of image coordinates manually

    y

    ZZrYYrXXr

    ZZrYYrXXrcyy

    x

    ZZrYYrXXr

    ZZrYYrXXrcxx

    )()()(

    )()()(

    )()()(

    )()()(

    033023013

    032022012

    0

    033023013

    031021011

    0

    ij

    j

    ij

    s

    l

    ss

    ll

    h

    Av

    v

    0

    0

    or

    X

    Y

    Z

    P (X, Y, Z)

    x

    y

    2 x

    Result: X, Y, Z coordinates

    3D measurements

    02.12.2014 GEPSUS Introduction 59

  • Principle:

    • Oriented and overlapping images are visualized using a Digital Photogrammetric Station (DPS)

    • Using special devices (shutter-glasses or polarized glasses), the human eye gets a 3D

    impression of the object, which can be used to measure 3D coordinates

    Advantages:

    • Low redundancy

    • Explicit measurements of significant and important object details

    • Low data volume

    • Post processing needless (blunder and errors free)

    • High accuracy

    Disadvantages:

    • High amount of manual work

    Commercial DPS packages:

    e.g. Leica Photogrammetric Suite (LPS), SocetSet, ZI-

    Image Station, etc.

    3D stereo measurements

    02.12.2014 GEPSUS Introduction 60

  • • Example stereoscopic analysis (Norway)

    Stereoscopic plotting

    02.12.2014 GEPSUS Introduction 61

    Hardware:

    • 2 workstations with 3

    moitors each: 2 x 27´´ 3D

    and 1 touch-screen monitor

    • Wire-less shutter glasses

  • DTMs and DSMs (LAS and GeoTIFF formats)

    Orthophotos (GeoTIFF format)

    GIS

    • Import geodata

    • Visualization

    • Overlapping

    • Processing

    • Analysis

    • Export

    Photogrammetric products for GEPSUS

    02.12.2014 GEPSUS Introduction 62

    Operation possible in GIS environment

    Reference system: UTM32N, WGS84

  • Webinar organization - calendar

    02.12.2014 GEPSUS Introduction 63

    Day Time H Topic Responsible Material

    2/12/2014 9:30-11:30 2 Introduction to aerial

    photogrammetry: instruments,

    calibration and processing workflow

    Terra Messflug PPT

    2/12/2014 12:30–14:30 2 Flight planning and image acquisition Terra Messflug Demo

    3/12/2014 9:30-11:30 2 Trajectory estimation, Ground survey

    and Aerial Triangulation

    Terra Messflug Demo

    3/12/2014

    12:30–14:30 2 Orthophoto and True-Orthophotos

    generation

    Terra Messflug PPT + Demo

    4/12/2014

    9:30-11:30 2 3D point cloud extraction and digital

    surface modelling

    Terra Messflug PPT + Demo

    4/12/2014

    12:30–14:30 2 3D point classification Terra Messflug PPT + Demo

    5/12/2014

    9:30-11:30 2 Case study using Trento dataset Graphitech Demo

  • • Combination of theory and live demonstrations

    • Demonstration with Trento dataset and examples from other projects

    • Proposed exercises with Trento dataset and Global Mapper http://www.bluemarblegeo.com/products/global-mapper-download.php

    • Skype platform

    Webinar organization - method

    02.12.2014 GEPSUS Introduction 64

    http://www.bluemarblegeo.com/products/global-mapper-download.phphttp://www.bluemarblegeo.com/products/global-mapper-download.phphttp://www.bluemarblegeo.com/products/global-mapper-download.phphttp://www.bluemarblegeo.com/products/global-mapper-download.phphttp://www.bluemarblegeo.com/products/global-mapper-download.phphttp://www.bluemarblegeo.com/products/global-mapper-download.php