Rectification on YZ plane. Rectification on XZ plane.

• Rectification on YZ plane

Rectification on XZ plane

• Rectification on XY plane

Detail of the rectified image on XY plane

ONE (OR TWO) VANISHING POINTS GEOMETRYONE (OR TWO) VANISHING POINTS GEOMETRY

The original image (Volubilis, Morok)

RECTIFICATION

Observation of two families of parallel lines

RECTIFIED IMAGE

The deformation on the upper line are due to lack of

planarity of the mosaic

CORRECT SCALING

• The Vanishing Point GeometryVanishing Point Geometry is a powerful tool.

+• With oneone Vanishing Point Vanishing Point only: the image can be rectified: the ratio base/height is not correct• With twotwo Vanishing Points Vanishing Points (one horizontal, one vertical): the image can be rectified: the ratio base/height is not correct• With threethree Vanishing Points Vanishing Points: the image can be rectified, the ratio base/height is correct, the orientation can be assessed, both for the original and the rectified image, apart from one unknown scale factor

-• Camera axis must be very inclined• Only WA are suitable

PhotoModelerPhotoModelerby Eos Systems - Canada

• A project:

• The inverse camera :

from the pictures to the recostruction

of the 3-D model

Determination camera parameters

Low-cost software 1000 $

8)

Selection measure unity

Options

Project:

Type of camera

List

resolution of the images

camera parameters

Image Selection in the project

Punti omologhi collimati in numero minimo

Options of the computation

Modello Wireframe

final model with texture

The inverse camera :

Unknown camera parameters

Processo vincolato

constraints to set

Format .cam

Tests

Sintetic image

= -20°

= 40°

Zc= 3.25

'

'

'

'

A

D

D

A

B

C

C

B

10955.0000.0

10757.0302.1

10239.0022.1

10000.0000.0

10279.0589.0

10116.0195.1

10181.1856.0

10776.0439.0

1010

1011

1001

1000

1110

1111

1101

1100

1000

273.00188.0423.0

064.00978.00

127.00088.0906.0

T

given

Dalla deviazione standard (SQM) si è dedotto che nelle collimazioni si ha un’incertezza di 0,6 pixel per unità di misura.

PhotoM

odeerP

hotoModeer

Dalla deviazione standard (SQM) si è dedotto che nelle collimazioni si ha un’incertezza di 0,5 pixel per unità di misura.

PP

FF

33

Cubo 100_30_f28

Results: Φ 29.935 deg Expected: 30 deg

  θ 9.953 deg   10 deg

  Κ -0.010 deg   0 deg

  f 27.957 mm   28 mm

  ppx 400.546 pxl   400 pxl

  ppy 299.985 pxl   300 pxl

Cubo 60_15_f35

Results: Φ 15.453 deg Expected: 15 deg

  θ -30.038 deg   -30 deg

  Κ -0.245 deg   0 deg

  f 35.001 mm   35 mm

  ppx 394.151 pxl   400 pxl

  ppy 297.443 pxl   300 pxl

PP

FF

33

Ancona – Traiano’s arch – Model formation by Photomodeler

Ancona – Traiano’s arch – The residuals of the Model formation

Ancona – Traiano’s arch – the camera stations

A practical example of 3D- A practical example of 3D- reconstructionreconstruction

The Ho Chi Minh’s house in The Ho Chi Minh’s house in

HanoiHanoi

Fro

ntF

ront

Lef

t sid

eL

eft s

ide

Rig

ht s

ide

Rig

ht s

ide

Bac

k si

eB

ack

sie

inverse Camera

Interior camera Parameters stored formato .cam

Some phases of the observation and referencing process

Points in 3D space Height of the parapet =

1 Module

PF3 measurementsPF3 measurements

PhotoModeler

Exterior

P F 3

Interior

Model Recostruction in AutoCad

PhotoModeler: utilized by non-specialised operators very powerfulVery good interface Many options

3D modeling VRML exportRectificationNon-metric imagesControl informationsNo need to supply approximate value

Only one control information is missing: the camera station coordinates

• ARPENTEURARPENTEUR

by Pierre Drap – Pierre Grussenmeyerby Pierre Drap – Pierre Grussenmeyer

• (ARchitectural PhotogrammEtry Network Tool for EdUcation and Research).

• Designed in 1998 by two research teams (GAMSAU-CNRS and ENSAIS-LERGEC)

• Mainly for Education

9)

• The running from the net• Thus it can be easily and freely used

– from anywhere, by anybody all over the world and – with whatever operating system.

• www.arpenteur.net.

• Dedicated to architectural photogrammetry and close range terrestrial photogrammetry, but aerial images (limited to a few Mb) can be also handled.

• Photos from a wide range of calibrated cameras – metric – non-metric camera, – digital amateur camera

• ARPENTEUR permits a better knowledge of basic photogrammetric techniques

• stereoplotting, • image correlation, • architectural photogrammetry • architectural modelling

• Output results can then be viewed as text-file, DGN, DXF, VRML file for a further processing with Internet or CAD systems (e.g.: MicroStation, AutoCad).

• There are today on the market many good products for any purpose and for any pocket

• The quality of a system is based not only in the accuracy but anso in the efficiency

• Photointerpretation

• Orthophoto

Conclusions:

Acknowledgements

Thanks to my students who helped me, part of this material is their graduation thesis:

• Matteo Cinti, Marco Battistelli• Paolo Clini• Gianluca Gagliardini• Stefano Benassi• Paolo Margione• Loretta Alessandroni• Ivan Catini• Floriano Capponi• Ingrid Luciani