parameters co directly used:
object space stage is usually
dramatically lo orientation sol
At the same reconstruction
afterwards. At the end of t
acquired along link between a
the particular photo-station:
photo-station axes that pass
itself; the tra frames acquir
homography t images:
Where H is coordinates re
K is the calibr interior orient
matrix betwee Since the rotat
the acquisition are known, th
between them tries to link th
other image: t between the
tracked for at points extracte
approach, the on one sequen
LSM, Gruen, At the end of
with all the operator is pe
3.2 DSM gen
The DTM gen at the Univer
Least Squares Multiphoto G
method Grue known local
transformation minimize the
between the p further extend
constraint of accommodate
redundancy an input data the
parameters of object surface
In most cases the image blo
normally dens Should it be n
omputed from in this case th
referencing of y very fast sinc
ow: just few ite lution.
time the algo n of the triplet
the image seque g the tunnel mu
adjacent strips geometry of th
approximately location, rotate
ses through the ansformation fr
red in that pos that depends o
ܪ ൌ the transform
epresenting the ration matrix of
tation paramete en the two imag
tion angle betw n device, once
he homography m can be easily
he tie points on that should lead
strips since th least 3 images
ed are not enou homologous po
nce are detected 1985.
f the procedure images and
erformed.
neration
neration program rsity of Parma
s Matching LS Geometrically
en et Baltsavia area-based me
n and an aff squared sum
patch image an d the method c
the collinearity any number of
nd providing gr program requi
f the images an to densify.
the seed point ock orientation
se and well d necessary, a pr
m the essential he outlier can b
f the triplets. A ce the outlier pe
erations are requ orithm already
ts that can be ence orientation
ust be tied toge are made extre
he images acqu y the acquisiti
es the camera a e perspective c
rom the image sition is well-a
on the rotation ൌ ܭܴܭ
ିଵ
mation matrix e homography b
f the camera, w ers, and R is t
ges. ween adjacent im
the interior or y approximating
computed. The n one image w
d to the highes he tie points e
s along the ima ugh however, u
oints correspond d by an image c
e a final bund with the GCP
m Dense Match since 2006, im
SM method G Constrained M
as, 1988. The ethod, where a
fine mapping m of the grey
nd the templat coupling LSM
y equation; the f images availab
reater reliability res interior and
nd seed points t ts may be taken
procedure of S distributed over
reliminary featu l matrix stage
e filtered durin Anyway the tri
ercentage tends uired to asses a
y provides a m easily concate
n stage all sequ ther. In this cas
emely easy than uired from the
ion device, at around an horiz
center of the ca e spaces of al
approximated b n angle betwee
using homoge between two im
which contains a the relative rot
mages is enforc ientation param
g the transform e SfM strategy,
with the other i st level of tight
extracted so fa age sequences.
using an Area B ding to the tie p
correlation algo le block adjust
P identified by
her being deve mplements bot
Gruen, 1985 an Matching MG
e former is a a linear radiom
are introduce values differ
te image. The with the geom
e method allow ble, exploiting i
y of the solutio d exterior orient
that approximat n from the outp
Section 3.1, tha r the image fo
ure based extra e are
ng the rifocal
s to be a good
metric enated
uences se the
nks to same
each zontal
amera ll the
by an en the
1 enous
mages, all the
tation ced by
meters mation
then, in the
tening ar are
If the Based
points orithm
tment y the
eloped th the
nd the GCM
well- metric
ed to rences
latter metric
ws to image
on. As ntation
te the put of
at are ormat.
action and
and t The
work Imag
effic such
accep contr
exten rang
3.3
At th simp
geolo are r
the comp
the comp
whic point
ortho whil
abutm circu
A so galle
first geom
matching is pe the epipolar con
matching proc kflow and impl
ges may unde iency and com
h as template ptance threshol
rol the processi nsively in DTM
e on cultural he
Figur
Orthophoto p
he end of the D plified to be a
ogists. To this a reprojected to g
tunnel. Since plex, a simplifi
mapping of it posed of two p
ch support a se t under the u
ogonally on th e the DSM p
ments are pro ular vault.
Figure 3. S oftware to prod
ery was proprie all the DSM p
metry: the dime erformed by us
nstraint to filter cedure is emb
ements a parall ergo to epipo
mputational spe window size,
ld, output point ing. Dense Mat
M generation of eritage artworks
re 2. DSM of a
roduction
DSM generation analyzed and
aim the images enerate an orth
the geometry ied model of th
ts intrados; in planar surfaces
mi-circular vau pper part of t
he planar part portion whose
ojected along
Simplified tunn duce the ortho
etary developed points are map
ension of the s sing again the S
r out possible m bedded in an i
lel dense match olar resamplin
eed; a number , image corre
nt spacing, etc. tcher has been u
f rock faces as w s with multi-ph
a tunnel section n the data colle
queried more s used during D
hophoto of the i of the tunne
he gallery is req n particular th
approximating ult see figure 3
the abutments of the simpli
e height is hi the radius dir
nel mapping geo ophoto of the i
d; its workflow pped on the sim
simplified mod SURF operator
mismatches. image pyramid
hing procedure. g to improve
of parameters lation method,
can be set to used and tested
well as in close oto coverage.
n. ected should be
easily by the DSM generation
inner surface of el is generally
quired to allow e geometry is
g the abutments 3: every DSM
are projected fied geometry,
gher than the rection of the
ometry intrados of the
w is as follow: mplified tunnel
del are derived r
d .
e s
, o
d e
e e
n f
y w
s s
M d
, e
e
e :
l d
XXII ISPRS Congress, 25 August – 01 September 2012, Melbourne, Australia
300
from the me abutments, ma
etc. and it’s o those of the D
projection is planes represe
circular surfac points, repres
generated ont simplified mo
length of proj Delauney tria
every pixel th computed; fin
sequence imag assigned to eac
In figure 4 a section is pres
Figur
4. 4.1 Rockscan