CLOSE-RANGE PHOTOGRAMMETRY WITH AMATEUR CAMERA

Dimitar Jechev

GIS Sofia Ltd., Bulgaria, BG-1000 Sofia, 5 Serdika Str., e-mail: jechev@gis-sofia.bg

Commission V, WG V/4

KEY WORDS: Close-range, Non-metric, Experiment, Digital, Adjustment, Building

ABSTRACT:

A frontage of a building has been experim

geodetic. The photogrammetric survey has been made by means of

station. The shortcomings of the amateur hardware, that have been used for the experiment, were c

entally surveyed and processed in two basically different methods — photogrammetric and

an amateur digital camera, and the geodetic survey - by a total

ompensated by application of a

specific mathematical model. The results from the two surveys performed have been compared. The RMS in the plane of the picture

(the building frontage) has been calculated to be +1 .9 cm and for the

Racurs Co. has been used for this experiment.

1. THEORY

1.1. Problems

The work with non-metric cameras for photogrammetric

purposes is accompanied by the following problems:

e Defining the image co-ordinate system (non-metric

cameras do not have fiducial marks).

e Defining the unknown elements of internal orientation

(focal length and image co-ordinates of the principle poin

of the photograph).

e Maintaining the elements of internal orientation unchanged

in time - usually when working with non-metric cameras,

the elements of internal orientation get slightly changed

after every single exposure.

e Defining the distortion of lens - the distortion with

amateur cameras often amounts to considerable values and

have substantial effect.

1.2. Solving the problems

There are three basically different methods for solving the

above mentioned problems known:

Calibration in advance. Before surveying, in a laboratory

(calibration centre) the unknown elements of internal

orientation and distortion of lens shall be defined. The

advantage of this method is that the calibration takes place

at a laboratory and hence better accuracy at defining of

unknown quantities is achieved. The problem with their

fluctuation in time remains.

Calibration during the processing. The unknown

elements are defined by means of a special mathematical

instrument. A larger number of control points is needed for

their defining - at least 5 points, and it is recommendable

8-10 points per model, compared to 3 points per model

when using metric camera.

Self-calibration. It is based on the mathematical means of

the geometry of overlapped areas for defining the unknown

elements. The principles involved are similar to the ones,

used for relative orientation of stereo-pair with analogue

instrument. It is specific for this method that it does not

require larger number of control points.

136

depth £6.1 cm. Photogrammetric software PHOTOMOD Lite of

1.3. Mathematical processing

In accordance with the co-linearity condition, every object

point, its image and the projection centre, should belong to one

and same line, called ray. Mathematically this can be

represented by means of the following 3 equations:

X; X X X, |

Y jm yg 4; R Y Y ( | )

0 f z Z, |

where: i=1 2 ..n

is the number of measured image points,

n

(s, y. 0) are image co-ordinates of point 1,

NT (t . .

Kos Vort ) are the elements of internal orientation,

À, is the scale factor,

R is the rotation matrix, defining the

spatial rotation of the geodetic co-

ordinate system in relation to the image

co-ordinate system. R is function of the

three angles @ , ¢, K,

(X, , y ? Z, y

(34.3.2, y are

projection centre O.

are the geodetic co-ordinates of point i,

the geodetic co-ordinates of

It is not justifiable to render different scale factor for each i

point. The scale factor could be eliminated by dividing the first

and second equations from the system by the third one:

HH (x —A m,,(Y, = Y), + m,(Z, - Z,)

P =x —x,+ i T ali

eS nl m,, (x, — X.) mar, -Y,)yrmad zz)

tt

T y m4 (X, — X,}+m, (Y, = Y) +my(Z, -Z,) E

REY TNT. m, (X, — X m,, (Y, — Y,)- m, -4 s

where: m are the elements of R matrix