SE422
Introduction
Photogrammetry dealswithderiving3dimensionalobjector
groundcoordinatesfromphotographicimageswithhighest
accuracyandwithouttouchingtheobject.
Theresultofphotogrammetric measurementsmaybe:
Numbers (coordinatesofobjectpointsin3Dcoordinatesystem)
Plots (mapsandplanswithplanimetric detailandcontourlines)
Images (mainlyorthopotos)
Someoftheadvantagesofphotogrammetry:
Noncontactpositioningmethod
Wecantakeimagesofalargenumberpointssimultaneously
Wealsocanarchivethephotographsfordocumentation.
Typesofstereoplotters
DirectOpticalprojectioninstruments:createa
truethreedimensionalstereomodel by
projectingtransparencyimagesthrough
projectorlenses.
Mechanicaloropticalmechanicalinstruments
cancreate3Dmodelthroughmechanical
means(spacerods)
Analyticalplotters:mostoftheworkdoneby
computer
DIRECTOPTICALPROJECTIONSTEREOPLOTTERS
1.Mainframe
2.Referencetable
3.Tracingtable
4.Platen
5.Guiderods
6.Projectors
7.Illuminationlamps
8.Diapositives
9.Levelingscrews
10.Projectorbar
11.Tracingpencil
MECHANICALPROJECTIONPLOTTER
Diapositives placedincarriersandilluminatedabove
Spacerodsarefreetorotateaboutgimbal jointsO
andO andcanslideupanddownthroughjoints
ModelairbasedefinedbyOO
Jointsfixedinpositionexcepttheirspacingcanbe
changed
Dr. Fahd Abanmy
ANALYTICALPLOTTERS
Developmentpossiblewithadvancesin
computers,digitalencodersandservosystems
Useexactmathematicalcalculationstodefine
stereomodel
Capabilities
Preciselymeasurexandyphotocoordinateson
bothphotosofastereopair
Accuratelymovetodefinedxandyphoto
locations
ANALYTICALPLOTTERS
Wild BC-1
Zeiss P3
Maincomponents
1. Thestereocomparator wherethetwoplates
areinserted
2. Acontrollercomputer
3. TheCADcomputer
4. Theoperatorsconsole
5. Servomotors
6. Encoders
Analogphotogrammetry suffersfromthelimitationof
theopticalmechanicalinstruments.
Itonlydealswithcentralperspectiveprojection
PanoramicandstripphotographicsystemsandXrayradiographyare
someexamplesthatcannotbeaccommodatedbyanaloginstruments
Difficulttoeliminatesomeerrors
Eliminatingerrorsduetoinfluenceslikethoseoflensdistortion,
atmosphericrefractionandfilmdeformationaredifficultor
impossibletoincorporateinopticalmechanicaldevicesbutare
accomplishedeasilythroughmathematicalmodelswithacomputer.
Difficulttoincorporateauxiliaryinformation
AuxiliaryinformationdatasuchasGPSdataorinertialnavigation
systemsdatamaynotbedirectlyenforcedbutmaybeconsideredas
adequatelyweightedparameterstoreinforceratherthanoverride the
geometricstrengthofnormalphotogrammetric procedures.
Restrictedtorangeoffocallength
Analyticalphotogrammetry canaccommodatephotographyfromany
focallength(orusedtwodifferentfocallengthstoformamodel
Opticalandmechanicalerrorsfromlightraysand
mechanicalspacerodsarenotintroduced
Theycancorrectformanycombinationofsystematic
errors(lensdistortion,shrinkage,refraction,earth
curvature)
Ineveryphaseoftheiroperation,theycantake
advantageofredundantobservationsand
incorporatethemethodofleastsquaresintothe
solutionoftheequations.
TheoryofOrientation
Theobjectiveoforientationistotransform
centrallyprojectedimagesintoathree
dimensionalmodel,whichwecanuseto
plotanorthogonalmap.
StepsofOrientation
Interiororientation
Relativeorientation
Absoluteorientation
ElementsofInteriorOrientation
ElementsofInteriorOrientation
1. Thecamerafocallengthf
2. Locationoftheprincipalpointwithrespectto
thefiducial marks(xo,yo)
3. Allthedistortionparameters
A. Correctionforfilmshrinkageandexpansion
B. Correctionforlensdistortion
dr =ko r+k1 r3 +k2 r5 +k3 r7 rc =r dr
whereristheradialdistancefromtheprincipalpoint
ko,k1,k2,k3 arecoefficientofthepolynomial
Themismatchisresolvedintotwo
components:
xparallax:afunctionofelevationandcanbe
eliminatedbyloweringorraisingtheprojection
plane
yparallax:itisadifferenceinphotocoordinates
inyaxes.
Itscausesare:
tiltinthephotographs
unequalflyingheights
improperorientation
yparallaxissystematicallyeliminatedby
rotatingandtranslatingoneorbothofthe
projectors.
Dr. Fahd Abanmy
PROJECTIONSYSTEMS
Torecreaterelativeangularrelationships
Projectorsmusthaverotationalandtranslational
movementcapabilities
6possibleforeachprojector
(omega) xrotationalsocalledroll
(phi) yrotationalsocalledpitch
(kappa) zrotationalsocalledyaw
Xtranslation
Ytranslation
Ztranslation
MOVEMENTOFPROJECTORS
The selection of which
orientation elements to use in
order to eliminate y-parallax is
govern by:
For a given point, clear the y-
parallax by that orientation element
which cause a maximum y-
displacement at the point
At any subsequent point, use the
orientation element which will not
introduce y-parallax at a previously
cleared point.
Dr. Fahd Abanmy
INDEPENDENTMETHODOF
RELATIVEORIENTATION
1.Clearpy atpoint1with
2.Clearpy atpoint2with
3.Clearpy atpoint3with
4.Clearpy atpoint4with
5.Observepy atpoint5andovercorrectbyintroducing
150%correctiontothemeasuredparallaxwith
6.Repeatsteps15untilnopy existsatpoint5
7.Checkforpy atpoint6.
IndependentRelativeOrientation
Determine l, l, r, r, r.
DependentRelativeOrientation
AbsoluteOrientation
AbsoluteOrientation:
MathematicalModel
Acontrolpoint,withknowngroundandmodelcoordinates,
yieldsthreeequations.
Iffullcontrolpointsareused,aminimumofthreecontrol
pointsshouldbeavailabletosolveforthesevenparameters
oftheabsoluteorientation.
Thesepointsshouldnotbecollinear.
Therollangleacrossthelinedefinedbythecontrolpointscannotbe
determined.
Fiducial Marks
Fiducial marksaresmalltargetsonthebodyof
metriccameras.
Theirpositionsrelativetothecamerabodyare
knownthroughacalibrationprocedure.
Theydefinetheimagecoordinatesystem.
Inthatsystem,thepositionoftheperspective
centreisknown.
FormanddistributionofFiducial marksdepend
onthecameramanufacturer.
xx
yy
ImageCoordinateMeasurements
y
Y
xp p
Xp
yp
Yp
X
Comparator coordinates (x`,y`) Image coordinates (x,y)
Dr. Fahd Abanmy
ImageCoordinateMeasurements
y
Y
Xp
p
xp yp
x
Yp
X
Comparator coordinates (x`,y`) Image coordinates (x,y)
Dr. Fahd Abanmy
ImageCoordinateMeasurements
Wecannotdirectlymeasuretheimage
coordinatesoffeaturesofinterest.
Weusemachines(knownascomparators)to
measurethecoordinatesrelativetothe
machinecoordinatesystem.
Output:Machine/comparatorcoordinates.
Themachine/comparatorcoordinatesare
reducedtoimagecoordinates(i.e.,relativeto
theimagecoordinatesystem).
GeocentricUniversalSystem
Thissystemisextensivelyutilizedforphototriangulation,
particularlywhenalargeportionoftheearthssurfaceisinvolved.
Thissystemisextensivelyutilizedforphototriangulation,
particularlywhenalargeportionoftheearthssurfaceisinvolved.
Itisanorthogonalsystemwhoseoriginisatthecenteroftheearth.
TheXYplanecontainstheequator.
TheXaxispassesthroughthezero(Greenwich)meridian,
TheZaxispassesthroughtheNorthPole.
TheYaxiscoincideswiththe90o eastlongitudeinrighthanded
system.
GeodeticCoordinateSystem
Thesystemusesarotationangels(Latitude,
LongitudeandHeightaboveaparticular
ellipsoid(,,h)
Thegeodeticsystemisnotusedinanalytical
photogrammetry directlyduetotheinherent
inconvenienceofhavingtoexpressthevalues
innonlinearunits.
Thesearegenerallyconvertedintooneofthe
rectangularsystems.
LocalCoordinateSystem
Thissystemhasthreeadvantages
1. Becauseitislocal,oneneednottocarrytoo
manydigits.
2. Thezaxisparallelsthelocalvertical,which
furthersimplifiestheworkpermittinga
separationofplanimetry fromelevation
values.
3. Thesystemisreadilyrelatedtothedirection
ofnorth,eastandgravityinlocalproject
area.
Dr. Fahd Abanmy
CoordinatesTransformation
Itsoftennecessarytoestablishthe
relationshipbetweentwocoordinatesystems
inordertotransferanumberofpointsfrom
onetotheother.
Theprocedurerequiresthatsomepointshave
theircoordinatesknowninbothcoordinate
systems.
Thesepointsarecalledcontrolpoints.
Theeffectoftransformationofabodycanbe
asimplechangeinlocationandattitude(without
anychangeinshapeandsize)
acomplexone(changeinbothshapeandsize)
Somethinginbetween(variationinsizewithno
changeinshapeorattitudeorlocation.
Y
Y
X TX
TY
2Scales X
X
All
Y Y Y
TX
TY
Scale X X X
Rotation Translation
TwoDimensionalConformal
Transformation
2Dmeansthatthecoordinatesystemslieona
planesurface.
Aconformaltransformationisoneinwhichtrue
shapeispreservedaftertransformation
Itisnecessarythatcoordinatesofatleasttwo
pointsbeknowninbothcoordinatesystems.
Accuracyinthetransformationisimprovedby
choosingthetwopointsasfarapartaspossible.
Ifmorethan2controlpointsareavailable,use
theleastsquaresmethod.
ya =s ya
X
xb =s xb Y
yb =s yb
b
Rotation
y
Y a
yb sin
yb cos
x
xb sin
X
X cos
X =x cos y sin
Y =x sin +y cos
Dr. Fahd Abanmy
N
N b
Translation TE a
X=X +TX TN E
Y=Y +TY
E
Combiningthesesteps
X=s x cos s ysin +TX
Y=s x sin +s ycos +TY
Or
wherea=scos
X=a x b y+c b=ssin
Y=b x +a y +d c=TX
d=TY
Inmatrixform
2n A4 4 X 1 = 2 n L1 + 2 nV1
a
yi b v x
x 1 0 X
A= i X = L = i V = i
yi xi 0 1 c Yi v yi
d
a 74.794 12.202
b -67.123 53.432
Dr. Fahd Abanmy
General2DConformalTransformation,
Example
4-Parameter Coordinate Transformation Program
______________________________________________________________________________________
Solution
Forming the B-matrix and f-matrix:
x1 y1 1 0 X1
y1 x1 0 1 Y1
x y2 1 0
X
2 2
y2 x2 0 1 Y2
B := f :=
x3 y3 1 0
X3
y3 x3 0 1 Y3
x4 y4 1 0 X4
y4 x4 0 1 Y4
Dr. Fahd Abanmy
General2DConformalTransformation,Example
(
N := B B
T ) 1
The variance-covariance matrix is: QXX := N
T
t := B f
102157.371
t=
1161.611
0.018
0.001
Dr. Fahd Abanmy
General2DConformalTransformation,Example
:= N t 0.99977
=
0.01137 The resisuals are V := B f
0.00211
0.01222
0.002
0.013
0.004
0.019
V=
0.002
0.020
0.004
Dr. Fahd Abanmy 0.013
General2DConformalTransformation,Example
Xa := 1 xa + 2 y a + 3 Xa = 74.913
Ya := 2 xa + 1 y a + 4 Ya = 11.361
Xb := 1 xb + 2 y b + 3 Xb = 66.502
Yb := 2 xb + 1 y b + 4 Yb = 54.195
Dr. Fahd Abanmy
2DAnetransformation
Twomaindifferences:
Differentscalefactorsinxandy
Compensatefornonorthognality oftheaxissystem.
Theuseoftheaffinetransformationisappliedto:
Transformcomparatortophotocoordinates(film
distortion)
Stereomodelconnections
Modeltosurveycoordinatestransformation.
Scalechange
x'=sx x
y'=sy y
Correctionfornonorthognality y
y
x"=x
y' xtan
y" = x' tan p
cos
y'
cos y
x"= x
y'
y" = x' tan
cos
y'
X = x' cos x' tan sin + TX
cos
sy y
X = s x x cos s x x tan sin + T X
cos
sy y
Y = s x x sin + s x x tan cos + T y
cos Dr.
Fahd Abanmy
Combiningthesefoursteps:
sy y
X = s x x cos s x x tan sin + T X
cos
sy y
Y = s x x sin + s x x tan cos + T y
cos
Since tan =
sin
cos
cos cos + sin sin sin
X = TX + s x x sy y
cos cos
cos sin sin cos sin
Y = TX + s x x + sy y
cos cos
substituting
a0 a1 a
cos( ) 2 sin
X = TX + s x x sy y
cos b cos
b0 b1 2
sin( ) sin
Y = TX s x x + sy y
cos cos
Dr. Fahd Abanmy
Thefollowingsubstitutionsaremade:
a0 = TX b0 = TY
cos( ) sin( )
a1 = s x b1 = s x
cos cos
sin cos
a2 = s y b2 = s y
cos cos
The final form of the affine transformation:
X = a0 + a1 x + a2 y
Y = b0 + b1 x + b2 y
Dr. Fahd Abanmy
a2 b1
= tan 1 = tan 1
b2 a1
cos cos
s x = a1 s y = b2
cos( ) cos
TX = a0 TY = b0
2n A6 6 X 1 = 2 n L1 + 2 nV1
a0
a
1
x yi 1 0 0 0 a X v X
A= i X = 2 L = i V = i
0 0 0 xi yi 1 b0 Xi vYi
b1
b2
3pointsgiveuniquesolution
Morethan3points useleastsquaressolution
Dr. Fahd Abanmy
a 74.794 12.202
Input Values: Note that lower case values represent observed comparator coordinates while the upper
case represents the known camera calibration coordinates for the respective fiducial values
_____________________________________________________________________________________
(
ATTAB
N := B ) 1
Dr. Fahd Abanmy
General Affine Transformation, Example
51079.018
583.52
0.018 The solution vector is: X := N t
t=
578.092
51078.353
0.001 0.99977
0.01134
0.00211
=
0.01140
0.99977
Dr. Fahd Abanmy
0.01222
General Affine Transformation, Example
The resisuals are V := AB X f
0.001
0.016
0.001
0.016
V=
0.001
0.016
0.001
0.016
The reference variance for the adjustment is
T
V V
:= = ( 0.001)
Dr. Fahd Abanmy
2
Xa = a1 xa + a2 ya + a0 Xa=74.913
Y a = b1 x a + b2 y a + b0 Xa=11.359
Xb = a1 xb + a2 yb + a0 Xa=-66.504
Y b = b1 x b + b2 y b + b0 Xa=54.197
xyzXYZ
Thethreedimensionaltransformationinvolvingno
changeinshapeissymbolizedby:
X = sM T X + T
Where
X =[XYZ]T coordinatesaftertransformation
X =[xyz]T coordinatesbeforetransformation
T=[Tx Ty Tz]T vectorofthreetranslation
factors
M=anorthogonalthreeangelrotationmatrix
s=scalefactor
z c
y1 = y ' cos + z ' sin y c o
s
os
z1 = y ' sin + z ' cos
y1
1 0 0
M = 0 cos sin
y
y sin
0 sin cos
Thesecondaryrotation,theangle()about
they1axisandx1y1z1 systemintox2y2z2
system.ThecoordinatesofpointPinthis
twicerotatedsystem:
x2 = z1 sin + x1 cos z1
z2
y2 = y1
z 2 = z1 cos + x1 sin
cos 0 sin
M = 0 1 0 x1
sin 0 cos x2
x = x 2 cos + y 2 sin y
y = x 2 sin + y 2 cos
z = z2
cos sin 0 x
M = sin cos 0
0 0 1 x2
Inviewofmathematicalprocedures,i.e.the
orderofmatrixmultiplication,itisbetter
writtenasfollows:
m11 m12 m13
M = M M M = m21 m22 m23
m31 m32 m33
cos cos sin sin cos + cos sin cos sin cos + sin sin
M = cos sin sin sin sin + cos cos cos sin sin + sin cos
sin sin cos cos cos
x
x
ThematrixMiscalledtherotationmatrix.
Eachelementofthematrixisthedirection
cosineswhichrelatesthetwoaxissystems,
andcanbeexpressedintermofdirection
cosinesas:
cos xx' cos xy ' cos xz '
M = cos yx' cos yy ' cos yz ' M= (i,j,k)
cos zx' cos zy ' cos zz '
2. Thesumoftheproductoftwodifferentrowsor
columnsoftheMmatrixisequalto0.
iTj =iTk =jTk =0
m11m21 +m12m22 +m13m23 =0
m11m31 +m12m32 +m13m33 =0
m21m31 +m22m32 +m23m33 =0
x
x
Step2 ScalingandTranslation
Itisnecessarytomultiplyeachequationbyascale
factorandaddthetranslationfactorsTx,Ty,Tz.
X=sx +Tx =s(m11x+m21y+m31z)+Tx
Y=sy +Ty =s(m12x+m22y+m32z)+Ty
Z=sx +Tx =s(m13x+m23y+m33z)+Tz
Inmatrixform =sMTX +T
Unknowns=,,,s,T X
x,Ty,andTz
Uniquesolution:2horizontalpoints+3vertical
pointsareknowninbothsystems
Theseequationsarenonlinearequationsinvolving7
unknowns.
Aspatialsimilaritytransformationrequiresatleasttwo
planimetric pointsandthreeheightcontrolpoints,ortwo
fullcontrolpointsandaheightcontrolpointnotinthe
sameline.
Ifthereareredundantmeasurementsaleastsquares
solutionwillbenecessary.
Dr. Fahd Abanmy
Tosolvetheseequations,theyarelinearized usingTaylors
theorem.
X=f(s,,,,Tx,Ty,Tz)
F F F F F F
dTx + dT y + F
X Fo +
ds +
d +
d + d + T T T dTz
s o o o o x o y o z o
X FXo + a11 ds + a12 d + a13 d + a14 d + a15 dTx + a16 dT y + a17 dTz
Y FYo + a 21 ds + a 22 d + a 23 d + a 24 d + a 25 dTx + a 26 dT y + a 27 dTz
Z FZo + a31 ds + a32 d + a33 d + a34 d + a35 dTx + a36 dT y + a37 dTz
Wherea11=m11 x+m21 y+m31 z
a12 =0
:
:
Dr. Fahd Abanmy
Fo evaluated at the initial approximations for the 7 unkowns
F
,etc.arepartialderivativeofthefunctionFwith
s o
respecttotheindicatedunknownsevaluatedatthe
initialapproximations.
ds,d,d,etc.areunknowncorrectionstobeapplied
totheinitialapproximations.
so + ds = s
o + d =
o + d =
:
Tzo + dTz = Tz
First approximations:
==0
xb xa X XA
= , = B , o =
yb y a YB YA
( X B X A )2 + (YB YA )2
so =
(xb xa )2 + ( yb ya )2
b
xa= so xa cos o so ya sin o
ya= so ya sin o + so xa cos o Y
za= so za y
Txo= XA xa a x
Tyo= YA ya
Tzo= ZA za X
Dr. Fahd Abanmy
Thecollinearity conditionequations
Collinearity is the
condition in which the
exposure station of any
photograph, an object
point, and its photo
image all lie on a straight
line.
Collinearity ConditionEquations
Let:
Coordinates of exposure station be XL, YL, ZL
Coordinates of object point A be XA, YA, ZA
Coordinates of image point a of object point
A be xa, ya, za wrt xy photo coordinate system
(of which the principal point o is the origin;
correction compensation for it is applied later)
Coordinates of image point a be xa, ya, za
in a rotated image plane xyz which is parallel
to the object coordinate system
Transformation of (xa, ya, za) to (xa, ya, za)
is accomplished using rotation equations,
which we derive next.
z c
x1 = x y c o
so
y1 = y cos + z sin
z1 = -ysin + z cos y1
Similarly, we obtain equations for phi
rotation about y axis:
x2 = -z1sin + x1 cos y
y 2 = y1 y sin
z2 = z1 cos + x1 sin
And equations for kappa rotation about z
axis:
x = x2 cos + y2 sin
y = -x2 sin + y2 cos Dr. Fahd Abanmy
z = z2
FinalRotationEquations
We substitute the equations at each stage to get the following:
x = m11 x + m12 y + m13 z
y = m21 x + m22 y + m23 z
z = m31 x + m32 y + m33 z
In matrix form: X = M X
where
x m11 m12 m13 x'
X = y M = m21 m22 m23 X ' = y '
z m31 m32 m33 z '
ReviewofCollinearity Equations
Collinearity equations:
Collinearity equations:
m ( X X L ) + m12 (YA YL ) + m13 ( Z A Z L )
xa = xo f 11 A are nonlinear and
m31 ( X A X L ) + m32 (YA YL ) + m33 ( Z A Z L )
involve 9 unknowns:
m ( X X L ) + m22 (YA YL ) + m23 ( Z A Z L )
ya = yo f 21 A
m31 ( X A X L ) + m32 (YA YL ) + m33 ( Z A Z L ) 1. omega, phi, kappa
Where, inherent in the ms
xa, ya are the photo coordinates of image point a 2. Object coordinates (XA,
XA, YA, ZA are object space coordinates of YA, ZA )
object/ground point A 3. Exposure station
XL, YL, ZL are object space coordinates of exposure coordinates (XL, YL, ZL )
station location
2 equations for every point in
f is the camera focal length every photo.
xo, yo are the offsets of the principal point coordinates
ms are functions of rotation angles omega, phi, kappa
(as
derived earlier) Dr. Fahd Abanmy
LinearizationofCollinearity Equations
Rewriting the collinearity equations:
r where
F = xo f = xa
q q = m31 ( X A X L ) + m32 (YA YL ) + m33 ( Z A Z L )
Linearized
F
Collinearity
F F
EquationsTerms
F F
F0 + d + d + d + dX L + dY L
0 0 0 X L 0 YL 0
F F F F
+ dZ L + dX A + dY A + dZ A = x a
Z L 0 X A 0 Y A 0 Z A 0
G G G G G
G0 + d + d + d + dX L + dY L
0 0 0 X L 0 YL 0
G G G G
+ dZ L + dX A + dY A + dZ A = y a
Z L 0 X A 0 Y A 0 Z A 0
where
F0, G0: functions of F and G evaluated at the initial approximations for the 9
unknowns;
are partial derivatives of F and G wrt
F F G G
, ,
, , etc.,
the indicated unknowns evaluated at
0 0 0 0
the initial approximation
d , d , d , etc., are unknown corrections to be applied to the initial
approximations. (angles are in radians)
Dr. Fahd Abanmy
SimplifiedLinearized Collinearity Equations
Since photo coordinates xa and ya are measured values, if the equations
are to be used in a least squares solution, residual terms must be included
to make the equations consistent.
The following simplified forms of the linearized collinearity equations
include these residuals:
b11d + b12 d + b13d b14 dX L b15 dYL b16 dZ L + b14 dX A + b15 dYA + b16 dZ A = J + v x a
b21d + b22 d + b23 d b24 dX L b25 dYL b26 dZ L + b24 dX A + b25 dYA + b26 dZ A = K + v ya
where J = xa F0, K = ya - G0 and the bs are coefficients equal to the partial derivatives
In linearization using Taylors series, higher order terms are ignored,
hence these equations are approximations.
They are solved iteratively, until the magnitudes of corrections to initial
approximations become negligible.
Dr. Fahd Abanmy
2m An n X 1 = 2 m L1 + 2 mV1
m =numberofpointsineachphoto
n =numberofunknowns
A =thematrixofbs
X =thematrixofunknowncorrectionstotheinitial
approximations
L =matrixofconstanttermsJ andK
V =matrixofresidualerrorsinthemeasuredx andy photo
coordinates.
SpaceResectionbyCollineariy
Spaceresectionistheprocessinwhichthespatialposition
andorientationofaphotographisdeterminedbasedon
photographicmeasurementsoftheimagesofgroundcontrol
points.
Itinvolvesformulatingthecollinearity equationsforanumber
ofcontrolpointswhoseXP,YP,ZPgroundcoordinatesare
knownandwhoseimagesappearinthetiltedphotograph.
In matrix form:
2m A6 6 X 1 = 2 m L1 + 2 mV1
(b11 )1 (b12 )1 (b13 )1 ( b14 )1 ( b15 )1 ( b16 )1
(b ) (b 22 )1 (b 23 )1 ( b 24 )1 ( b 25 )1 ( b 26 )1
21 1
(b11 )2 (b12 )2 (b13 )2 ( b14 )2 ( b15 )2 ( b16 )2
A=
(b 21 )2 (b 22 )2 (b 23 )2 ( b 24 )2 ( b 25 )2 ( b 26 )2
(b11 )m (b12 )m (b13 )m ( b14 )m ( b15 )m ( b16 )m
(b 21 )m (b 22 )m (b 23 )m ( b 24 )m ( b 25 )m ( b 26 )m
d ( J )1 (v x )1
d (K ) (v )
1 y 1
d ( J )2 (v x )2
X = L= V =
(v y )2
dX L (K )2
dY L ( J )m (v x )m
dZ L ( J )m (v y )m
Dr. Fahd Abanmy
Eachcontrolpointgives2equations
Sincewehave6unknowns(d,d,d,dXL,
dYL,dZL),thenweneedatleast3control
points.
Thesecontrolpointsshouldnotbeonornear
straightline.
Morethan3points,solutioncanbeobtained
bythemethodofleastsquares.
Solutionbythecollinearity equations
1. Determineasetofinitialvaluesfortheunknown
quantities(XLo,YLo,ZLo,o,o,o)
2. DeterminethevaluesofJ,K,andallbs
3. Solvethelinearized equationsforthedifferential
corrections
4. Addthesecorrectionstotheinitial
approximationvalues.
5. Iteratethesolutionoftheseequations(step24)
untiltheysatisfythedesiredaccuracy.
Dr. Fahd Abanmy
Initialvalues:
ZLo : usealtimeterreadingorthescaleequation:
AB
H= f + h AB
ab
XLo andYLo :useaverageXandYcoordinatesofthe
groundcontrolpoints.(oruse2Dconformaltofindthem)
o,o :canbeassumedaszero(nearverticalphotograph)
o :maybeestimatedbymeasuringtheangle
fromgroundnorthtothepositiveyphotoaxis.
xb xa XB XA b
= , = , o =
yb y a YB YA Y
Acounterclockwiseangleis y
consideredpositive. x
a
Dr. Fahd Abanmy
X
HW4
Anearverticalaerialphotographtakenwitha152.916mmfocal
lengthcameracontainsimageofthreegroundcontrolpointsA
throughC.Refinedphotocoordinatesarelistedinthefollowing
table.
photo coordinates Ground coordinates
Point
x (mm) y (mm) X (m) Y (m) Elev.(m)
A -100.916 92.582 732.181 545.344 22.299
Answerthefollowingquestions:
1. WhatistheinitialapproximationflyingheightZLo?
2. WhataretheinitialapproximationsgroundcoordinatesXLo andYLo of
theexposurestation?
3. Whataretheinitialapproximationsangles and ?
4. Whatistheinitialapproximation angle?
5. Computetheexteriororientationofthephotographbyresection
usingthecollinearity equations.
Dr. Fahd Abanmy
Spaceintersection
Thecorrespondingrays
tothesameobjectpoint
fromtwooverlapping
photosmustintersectat
thepoint.
ItisusedtodetermineX,
Y,andZground
coordinatesofnew
pointsappearinthe
overlaparea.
Knownarethesixelementofexterior
orientationofthetwooverlappingphotos
UnknowngroundcoordinatesofnewpointA
(XA,YA,ZA)
Thelinearized formsofthespaceintersection
equationsforpointAare
4 A3 3 X 1 = 4 L1 + 4V1
(b14 )1 (b15 )1 (b16 )1
(b ) (b 25 )1 (b 26 )1
A=
24 1
Solutionbythecollinearity equations
1. Determineasetofinitialvaluesfortheunknownquantities
(XAo,YAo,ZAo)
2. DeterminethevaluesofJ,K,andallbs
3. Solvethelinearized equationsforthedifferentialcorrections
4. Addthesecorrectionstotheinitialapproximationvalues.
5. Iteratethesolutionoftheseequations(step24)untilthey
satisfythedesiredaccuracy.
pa
XAo andYAo:usetheparallax
equations:
x y
X A = X L + B a1 , YA = YL + B a1
o o
pa pa
where
hA istheelevationofpointAabovedatum
Histheflyingheightabovedatum
Bistheairbase(distancebetweenthe
exposurestations)
B = [ (XL2-XL1)2 + (YL2-YL1)2 ]1/2
fisthefocallengthofthecamera
pa istheparallaxofpointA(pa=xaxa)
HW5Spaceintersection
Orienation ofastereopair ofaerialphotographstakenwitha
153.819mmfocallengthcameraresultedintheexterior
orientationvalueslistedinthefollowingtable:
Left photo Right photo
XL (mm) 0.000 99.070
YL (mm) 0.000 -2.1785
ZL (mm) 152.819 154.325
Omega ( ) 0.0000 1.9153
Phi ( ) 0.0000 -3.265
Kappa ( ) 0.0000 -1.7248
Ifrefinedphotocoordinates foraparticularpointonphoto1are
x1 =68.671mmandy1 =17.768mmandthoseforthesame
pointonphoto2arex2 =30.699mmandy2 =14.565mm,
computetheobjectspacecoordinatesforthepointbyspace
intersection. Dr. Fahd Abanmy
RelativeOrientationofStereopairs by
Collinearity
Itisanumericalduplicationoftheoneprojector
methodofstereoplotter relativeorientation.
Theleftphotoisfixedinposition,itsorientationis
alsofixed,andmodelscaleisinitiallysetbyassigning
anarbitrarymodelbase.
Therightphotoisthenadjustedbyapplying
rotationsandtranslationsuntilallyparallaxis
cleared.
1 =1 =1 =XL1 =YL1 =0
ZL1=actualflyingheightand
XL1 =actualairbase=>This
fixesthescaleofthe
mathematicalstereomodel
approximatelyequaltothe
truescale.
ORZL1=0andXL1=photo
base=>thisestablishes
modelscalewhichis
approximatelyequalto
photoscale.
Initialapproximations:
Assumeverticalphotograph,(2 =2=2=YL2=0)
ZL2 =ZL1
Objectspacecoordinatesmaybecalculatedusingparallax
equations.
Thesolutionisiterateduntilallcorrectionsbecomenegligible
insize.
Themodelcoordinatesystemobtainedisarbitrarywithits
axesparalleltotheleftphotoaxissystemanditsorigina
distanceZL1 verticallybeneaththeleftexposurestation.
Modelcoordinatesofanynewpointscanbeobtainedusing
thespaceintersectionmethod.
Theseadditionpointscouldalsohavebeenusedinrelative
orientation,therebyaddingredundancytothesolution.
(K c )1
dZ A
dX v yc
24 L1 = 24V1 =
= 1
X
23 1
B M M
dYB
dZ B
M M M
M
M
(J f )1
dX F
(K f )
vx f ( )
1
dY
F 1 vy
f
( )
1
dZ F
Dr. Fahd Abanmy
MatrixForm
ForSixpoint 24 A23 23 X 1 = 24 L1 + 24V1
A errelativeorientationisdone,a3Dmodel
oftheobjectisformedwithanarbitrary
coordinatesystem.
AbsoluteOrientation
Acontrolpoint,withknowngroundandmodel
coordinates,yieldsthreeequations.
Aminimumofthreecontrolpointsshouldbe
availabletosolveforthesevenparametersofthe
absoluteorientation.
Thesepointsshouldnotbecollinear.
Selectionofpointsforanalytical
photogrammetry
Controlpoints
pointswhosegroundcoordinatesareavailablefromgeodetic
measurements(e.g.,GPS).
Theyareusedtodefinethedatumduringthebundleadjustment.
Checkpoints areusedtodeterminetheaccuracyof
coordinatesderivedbyphotogrammetric adjustment.
Theirobjectcoordinatesareavailableandhavebeenindependently
measuredbygeodeticmethods,however,theyarenotincludedinthe
photogrammetric adjustment.
Theyareonlyusedforcomparison.
Tie(pass)points areappliedtotietogetheroverlapping
photographsofablock.
Theymustbeidentifiedindifferentimages.
Theirgroundcoordinatesaredeterminedbyaerialtriangulation.
Insmallscalephotographs:(buildingcorner,single
trees,intersectionofroads)
Inlargescalephotographs:(smallstones,manholes)
Theyshouldbeselectedunderamirrorstereoscope
tomakesurethattheyarewelldefinedinall3
dimensionsandcanbemeasuredwell.
Signalizedpoints(Panelpoints)
Thesearepointswhichweremarkedbysignalsontheground
beforethephotographsaretaken.
Thesepointscanbeusedascontrolpointsiftheir
geodeticcoordinatesaremeasuredorsimplyastiepoint
foraerialtriangulation.
Theyshouldbedesignedwithagoodcolorcontrastwith
itsbackground.
Smallscale:aerialmapping(1:30,000) d=50to100cm
Largescale:cadastralphotogrammetry (1:4,000) d=7to13cm
Terrestrialapplication:(construction)1:250 d=4to8mm
Thesizeofthetargetshouldbelargerthanthemeasuringmark.
Artificialpoints
Thesepointsaremarkedintheemulsionofthefilmbypoint
transferdevices.
Itisnecessaryifmonocomparatorsareusedforimage
coordinatesmeasurement.
Usually,artificialpointscorrespondtowelldefinednatural
points.
Theaccuracyofmeasurementofartificialpointsisalmostthe
sameasthatobtainedwithsignalizedpoints.
Theaccuracyofbasiccontrolsurveysis
generallyhigherthanthatofsubsequent
photocontrolsurveys.
Theestablishmentofgoodgroundcontrolis
extremelyimportantaspectoftheoverall
photogrammetric mappingoperation.
Costofestablishinggroundcoordinatesfor
photogrammetric mappingcanbe20%of
totalmappingcost.
RequiredAccuracyandResults
Decide:
therequiredaccuracyofthesurvey
theequipmentneeded
thefieled techniquestobeused
TheUnitedStatesNationalMapAccuracy
Standardsgovernaccuracyrequirementsof
maps.
Accuracyinplanimetry
X = (X ) i
2
, Y = (Y ) i
2
n n
whereXiandYiarethecoordinatediscrepancies
betweenacceptedandmappositionsat
eachtestedpoint.
n isthetotalnumberofpointsusedfor
checking.
Theplanimetric standarderror(p)isexpressedby:
p2=X2 +Y2
Theaccuracyofthefinalmapdependson:
thescaleofandresolutionoftheaerial
photography
theflyingheightofthephotograph
thebase/heightratiooftheoverlapping
photograph
theaccuracyofthestereoplotting equipment
usedofthemeasurements.
Photographicendlapandsidelap
Endlap istheoverlappingof
successivephotosalongaflight
strip. GB
PE = x100
G
Sidelap istheoverlappingof
adjacentflightstrip
PSshouldabout30%
GW
PS = x100
G
Dr. Fahd Abanmy
LOSSOFSTEREOSCOPICCOVERAGE
Baseheightratio
B/H istheratiooftheairbaseofapairof
overlappingphotostoaverageflyingheight
aboveground.
ThelargertheB/H ratio,thegreaterthe
parallactic angles
LargeB/H ratiosdenots lowflyingheights
andlargexparallax:=>higheraccuracy
10 miles
FlightPlanning
FlyEastWesttominimize#ofstrips.
FlyingHeightaboveDatum
f
PhotoScale = 1 / 12,000 = ; H = (12,000)(6inches / 12inches / ft ) + 790 ft = 6790 ft
H havg
Squaregroundcoverageperphoto(dimension)(G)
9
Dr. Fahd Abanmy
FlightPlanning
Lateraladvanceperstrip(W).Thisisthe
distancebetweenflightlines.
FlightPlanning
Numberofflightlines.
Alignfirstandlastsothat0.3G(thesidelap amount)is
outsidenorthandsouthprojectboundaries.
Edge of coverage for first strip
0.3G 10 miles
0.2G = 1800 ft
Centerline of first strip
6.5 miles
FlightPlanning
Advanceperphoto(airbase=B).
0.6G (60% endlap)
0.4G = B = 3600 ft
FlightPlanning
Total#photosinmission.
NT = (19 photos / strip)x(6 strips) = 114 photos
Spacingofflightlinesonexistingmap.
Mapscale=1:24,000or1 =2,000
6144 ft / int erval
S map = = 3.07inches / int erval
2000 ft / in
First and last lines are 0.9 inches (1800 ft) inside
the project boundaries.
Terrestrialandcloserange
photogrammetry
Itdealswithphotographstakenwithcameras
locatedonthesurfaceoftheearth.
Thecameramaybehandheld,mountedontripods,
orsuspendedfromtowers.
Terrestrialphotogrammetry canbefurtherclassified:
ascloserangephotogrammetry ifthecameraobject
distanceissomewherebetween1:10mto100m,
asmacrophotogrammetry ifthecameraobjectdistanceis
inthe0.10to0.01mrange,
asmicrophotogrammetry whenthephotosareexposed
throughamicroscope.
Architectural&HistoricalPreservationapplicaitons
PoliceDepartmentapplications
Medicalapplications
facial pattern recognition
back pattern recognition
Approachesofphotographicanobject
Singleimages
Averycommonproblem
isthatweknowthe
shapeandattitudeofan
object'ssurfaceinspace
(digitalsurfacemodel)
butweareinterestedin
thedetailsonthis
surface(patterns,
texture,additional
points,etc.).Inthiscase
asingleimage
restitutioncanbe
appropriate.
original photo
Dr. Fahd Abanmy
rectified orthophoto
Stereographicprocessing
Thesameprincipleofaerial
photogrammetry
3Dmeasurementscanbeachieved
Resultsofstereographicrestitutioncan
be:
2Dplansofsinglefacades
3Dwireframeandsurfacemodels
listsofcoordinates
eventuallycomplementedbytheirtopology
(lines,surfaces,etc)
Examplesofdifferentimages,differentcameras,differentlenses(from
projectOttoburg,Innsbruck)tocombinewithinabundlesolution
CategoriesofTerrestrialCameras
MetricCameras
Nonmetric(Amateur)Cameras
Digitalcameras
MetricCameras
Stereometric Cameras Independent Metric Camera
Digitalcameras
Photographycanbetakenwithavarietyof
cameras;however,theresultmustbedigital
imagefiles.
Digitalcamerasworkthebestforschedule
andefficiency,withnolossofaccuracy.
Theresolutionofthecamerasdefinesthe
fieldprocedurestobeused,notthefinal
accuracy.
The()istheanglefromthecameraaxisto
horizontal
Depressionangle()ifthecameraaxisisinclined
downward.
Elevationangle()iftheaxisupward
Theangleofinclination()isanimportantvariable
forcertainelementarymethodsofdetermining
objectspacepositionsofpointswhoseimages
appearonoverlappingterrestrialphotos.
Coordinatesystem
Originatk
x axiscoincidewiththehorizon
yaxis passthroughoandn
f f
= t 90o
with elevation angle, the vertical
parallel will intersect at the zenith.
oz
= 90 tan 1
f
Computinghorizontalandverticalangles
Onceangle hasbeendetermined,horizontal
andverticalanglescanbecomputedfor
pointsonthephoto
ha' 1 xa
= tan 1 = tan
Lk hk f sec y a sin
aa'
= tan 1
( Lk hk ) sec a
ya cos
= tan 1
( f sec y a sin ) sec a
= a
' = ' 'a
" = 180 '
B sin ' B sin '
LA = L' A =
sin " sin "
X A = LA cos YA = LA sin
X A = B L' A cos ' YA = L' A sin '