QUALITY CONTROL PROCEDURE FOR PHOTOGRAMMETRIC DIGITAL MAPPING 
R.J. Ackermann 
International Institute for Aerospace Surveys and Earth Sciences (ITC) 
P.O. Box 6, 7500 AA ENSCHEDE, The Netherlands 
Tel. +31-53-4874392, Fax +31-53-4874335 
A. Eslami Rad 
National Cartographic Center of Iran 
Commission IV, Working Group 1 
KEY WORDS : Digital mapping, quality control parameters, positional accuracy, quality reporting 
ABSTRACT : 
Quality assurance has become a major concern in a digital mapping environment and quality control procedures 
have been implemented in photogrammetric production lines by many mapping organizations. There is yet some 
uncertainty about procedures, quality parameters, sampling and statistical methods to be applied due to lack of 
standards. In the present paper an attempt is made to describe a quality control system which can be applied to 
photogrammetric feature extraction. Quality control covers all process involved in the creation of a spatial database 
where data are mainly collected by photogrammetric technique. There is generally agreement on the main quality 
components like positional accuracy, attribute accuracy, completeness, logical consistency, lineage and time. Not 
all of them are equally important for quality assessment; this paper will focus on positional accuracy, semantic 
accuracy and completeness after data collection since they can be easily assessed through statistical quality control 
procedures. The most reliable method for final quality assessment is the field control; it is also the most expensive. 
For photogrammetric control the original photographs are generally used and therefore this method has its 
limitations; however it provides useful information for the subsequent field completion. Quality reporting is of prime 
interest to the user; part of the information can be transfered into the database in the form of a separate quality layer. 
1. INTRODUCTION 
The errors we have to deal with are of three types: 
Quality control can be defined as "the operational . gross errors and blunders : they do not belong to 
techniques and activities that are used to fulfil the sample of observations and must be 
requirements for quality" (ISO). Quality remains a eliminated. 
vague concept, hard to define and sometimes difficult . random errors: they are present in any set of 
to measure. Surveyors and photogrammetrists have observations and their characteristics are well 
always been concerned about the accuracy of their known, provided that the sample size is 
observations. sufficiently large (n=50) 
Today, the concept of quality has been put into a . Systematic errors : they occur because of many 
broader perspective and the development of GIS imperfections through the various processes. 
requires a better quality management. Technological Systematic errors introduce a bias in the 
development and new requirements from the users observations and affect the accuracy. 
had a strong impact on quality control and quality Extensive field control carried out at IGN (France) 
assurance. There are six fundamental quality has shown important local systematic errors 
components of spatial data (Thapa, Bossler, 1992) : (Grussenmeyer, 1994) 
lineage Classification accuracy tests need to be designed in 
positional accuracy order to judge whether a proportion of 
attribute accuracy misclassification in a given sample is acceptable or 
completeness not. 
logical consistency A binomial distribution can be used for testing the 
. temporal accuracy classification accuracy of topographic features. The 
Lineage describes the source material from which the acceptance sampling method deals with the optimum 
data were derived (photographs, control points) and number of ground samples N and an allowable 
the methods used for the production of the spatial number X of misclassifications (Ginervan, 1979). 
data. Positional accuracy can be assessed by This method gives an overall estimate of 
comparing a photogrammetric data set to an misclassification accuracy but does not differentiate 
independant reference data set of higher accuracy between omission errors (excluded from a class) and 
(e.g. GPS data or field survey data). commission errors (included in a class). Therefore, 
12 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996 
"A peat El ae Maa É—üuda a 
i TR Jm "ux WP ia mam al —B Quas "um ae —' n ma 
Ou as "AK Tim o -—- D gh. pup o) wA 
ne 
na