2 2 
dy =- b,y + bax + byxy - b,(2y? - 48?/3) * bg(x, - 282/3) + bg(x? - 2B?/3)y * byox(y^ - 2B^/3) 
+ by, (x2 - 28/3) (y2 - 28%/3) 
parameter set c proposed by El-Hakim and Faig: 
ParameLe. 5°" 
dx = ax + a,y + af 
dy = — a;y + ax + of, where 
A A 2 4 20082) + a;rlsin2) + agricosA + agrosin) + a r3cos3) + a r3sin3A 
q = a3rcos + a,rsin\ + a,r agr' cos 7 8 s 9 10 11 
r= x? + y? and À » arctan d» 
Parameter set f proposed by Grün: 
Pparaue ———— 
dx * a,y + a, xy + a4xy? + a,x?y + asy? + apx2y” 
2 
dy = b,y + byxy + baxy? + b,x?y + box? T box? y 
Parameter set k proposed by Kólbl and modified by Juhl: 
paramce-er re-—— 
dx = 84 3 (r? = r/x2) + a,x(sinf-1)2 + agg Sin(LT) + rle85 * X(agcosa * a;sina + agcos2a + agsin2a) 
dy = a,x + apy * a3 Ur? - r/rà) * a,y(sin E + as X sin GEL + rle85 $(agcosa + a,sina + agcosa 
where a = arctan X) + agsin2a), 
r, = a given constant (radial distance, where radial distortion is wanted to be zero for the 
second time) 
Parameter set 1 proposed by Mauelshagen: 
dx = a, xy + agy? + am + a;x%y + agxy” + agy> 
3 
dy = b, x + b„y + bax? + b, xy + box? T b,x’y + bgxy? + boy 
Parameter set m proposed by Schut: 
dx = c3xy + es y” + cjx?y + coxy” + 61 + 619^ 
dy = cıy + cox ^. c,x? + cexy + cgx?y + cox? + c1?x^y? + I 
Parameter set n proposed by Salmenpera and Kilpela: 
dx = b,x + bay + baxr? (1-r, /r) + b,xr^ (17r, /r) + boxrÓ (1-1, /r) + bg* 2xy + b, (r? + 2x2) 
dy --bjy * box * b4yr?(1-r,/r) 4 b,yr^ (17r /r) * bgxrf(-r /r) * be (x? + 2y?) + b; + 2xy, 
where To is a given constant (first radial distance, where radial distortion is wanted to be zero). 
These parameter sets were incorporated into Schut's bundle adjustment program /Schut 1978/. This 
program has been written to allow in a simple way the introduction of any choice and number of 
parameters. This has been achieved by dimensioning all affected arrays to variable length and by 
restricting the program to the use of one of many subroutines defining additional parameters. This 
subroutine is defined by its name and the number of parameters in the input to the program. For this 
investigation, the program has been used repeatedly with one of the parameter sets defined above at a 
time. 
RESULTS 
Photographs of the two testfields Jámijaárvi and Rheidt were used thus far. Each testfield was 
simultaneously flown with two wide-angle réseau cameras, a Vild RC8R and a Zeiss RMK-AR. The available 
six sets of photographs consist of a 3x3 block with 60Z overlap in both directions and of four 
photographs each covering the area of the block in one of the four different orientations. All 
photographs were measured on Zeiss PSKl stereocomparators used in the monocomparator mode. All 
testfield points were used as control points after the discovery that points not stereoscopically 
determined in the 3x3 block would have to be dropped unless they were used as control. 
Our stategy consists of three steps. 
- correct image deformation by means of the réseau, 
- determine and correct system distortion parameters and 
- carry out self-calibration. 
120