d: 
les 
ré- 
ion 
it- 
in- 
ng 
ite 
al. 
or- 
To 
1al 
or- 
on 
n- 
of 
ed 
he 
ür 
he 
of 
oO 
or 
he 
  
STANDARDIZATION OF EXPRESSIONS FOR ACCURACY, HALLERT 167 
Recommendations concerning a provisional terminology for some basic concepts in 
the theory of errors of photogrammetry. 
1. All information on accuracy in photogrammetry must be founded upon real measure- 
ments. 
2. The accuracy must be expressed with unique and well defined terms, Expressions as 
for instance (the measurements have been or can be performed with . . . .) 
an accuracy of . . . 
an uneertainty of . .. - 
an error of .. ., ete. 
should be avoided unless the definition of the expression in question is clearly stated in 
each individual case. 
3. Information on accuracy in photogrammetry should preferably refer to the fundamen- 
tal observations of image coordinates and parallaxes (coordinate differences). 
4. For the determination of accuracy three different procedures are considered. 
4i Direct measurements of known quantities. 
If the result M of a direct measurement of a certain quantity is compared with a 
more correct, given value G of the same quantity the error s of the measurement M is 
defined as 
c=M—G (1) 
The correction v to the measured value M in order to obtain the value G is defined as 
v = — = G—M (2) 
If G can be regarded as an errorless (true) value, at least in comparison with M, 
¢ is denoted a true error. If G is regarded to be a condition the error « may be interpreted 
as a discrepancy in the condition. 
The accuracy of direct measurements can under the made assumptions be determined 
as the positive square root of the mean square value of the true errors (discrepancies) 
or as the (root)-mean-square error according to the definition (see ref. l, p. 253 and 8, 
p. 12). 
sz [/— (3a) 
where n is the number of measurements. 
If the individual errors (discrepancies) are reduced with the arithmetic mean or in 
formula 
n 
the positive square root of the mean square value of the & is denoted standard érror s 
according to the formula (ref. 1, p. 252-253, and p. 277): 
[Ie &'] 
n 
Ss = (3b) 
The error according to (1) can be caused by large, regular (systematic) and irregular 
(accidental) errors in the measuring procedure (ref. 2, p. 1-3). Large (gross) errors are 
caused by carelessness or mistake and can usually be discovered and eliminated after 
simple checks. Regular of systematic errors affect the results of the measurements ac- 
cording to some fixed law or function (possibly unknown) concerning the size and direc- 
tion of the individual errors and often dependent upon the local circumstances. As soon 
as the law or function is known, such errors can be determined and consequently the er- 
ror can be corrected. A special case of this type of errors is the constant error affecting 
all measurements with equal size and direction.