product, therefore, actively pursuing quality improvements
through a continuous cycle.
According to ISO, the quality of a product or service is defined
as “the totality of characteristics of an entity that bear on its
ability to satisfy stated and implied needs" (ISO 8402, 1994).
Quality management is mostly developing in construction and
manufacturing industries. Many experts contributed to the
success of the quality movement, mainly American and
Japanese. Some of the internationally acknowledged names
have to be mentioned here: W. Edwards Deming, Joseph M.
Juran, Phillip B. Crosby, A. V. Feigenbaum, K. Ishikawa, and
G. Taguchi. They introduced different fundamental doctrines on
how to implement and improve quality.
The term total quality management is most often used as the
generic term that means the vast collection of philosophies,
concepts, methods, and tools now being used throughout the
world to manage quality (Godfrey, 1999). The initial ISO 9000
series, that practically applied the concept of total quality
management to all generic product categories, was published in
1987. Since than, additional standards have been published.
The ISO 9000 family now contains a variety of standards
supplementary to the original series. Revision of the basic ISO
9000 series were published in 1994 (known as ISO 9000 :
1994), and the most recently in 2000 (known as ISO 9000 :
2000).
There are two relevant international standardization
organizations, dealing, among many other topics, with
Geographic Information: International Organization for
Standardization (ISO) and European Committee for
Standardisation - Comité Européen de Normalisation (CEN).
The ISO/TC 211 is devoted to Geographic information /
Geomatics. From many topics of the TC 211 program, two are
directly addressing quality: Geographic information — Quality
principles (ISO/DIS 19113) and Geographic information —
Quality evaluation procedures (ISO/DIS 19114). The CEN TC
287 is devoted to geographic information. A prestandard, ENV
12656: “Geographic information — Data description — Quality”,
is available. Indirectly connected with data quality on
informative level is the prestandard ENV 12657: Geographic
information — Data description — Metadata.
Statistical techniques play an important role in quality
evaluation procedures. Generally, data are elementary for
quality control. Data could be unreliable due to not suitable
sample methods, way of selecting samples, measurement
methods and analysis methods. Ishikawa (Ishikawa, 1985)
stated: “If you get data which were acquired with measurement
instruments you can always doubt about its correctness.” Thus,
the knowledge and appropriate use of data processing methods
and statistics is the next important issue when dealing with
spatial data quality.
3. EVALUATION PROCEDURES OF SPATIAL DATA
QUALITY
The measurement of quality is a complex operation as there are
numerous definitions of quality itself. Description of quality is
consequently complex. Defining the components or elements of
spatial data quality is the very first step to provide relevant
quality information for the users. In addition to this, easily
understood measures of spatial data quality must be defined as
138
well as quality evaluation procedures and instructions for
transparent reporting.
Many people equate accuracy of spatial data with quality of
spatial data but in fact accuracy is just one component of
quality. Thus, it is necessary to widen the knowledge of spatial
data quality far more than taking into account only positional
accuracy known from analogue era.
The basic spatial data quality components defined in the current
international standards (ISO, CEN), are completeness, logical
consistency, positional accuracy, temporal accuracy and
thematic (or semantic — in CEN) accuracy. Additional general
information are lineage, usage, and purpose - in ISO or
homogeneity — in CEN. The terminology in both standards is
different, but there is no real difference in the context. It is often
difficult to refer a specific quality problem to specific quality
component, as they are many relationships between these
components. Further on, measures and evaluation methods are
not yet sufficiently specified for practical use.
A data quality evaluation procedure is accomplished through
the application of one or more data quality evaluation methods.
Data quality evaluation methods are divided into two main
classes, direct and indirect. Direct methods determine data
quality through the comparison of the data with internal and/or
external reference information. Indirect methods infer or
estimate data quality using information on the data such as
lineage. The direct evaluation methods are further subclassified
by the source of the information needed to perform the
evaluation. For both external and internal evaluation methods,
there are two considerations:
e automated or non-automated, and,
e full inspection or sampling.
Different strategies for spatial data quality control could be
used. The producer of the dataset should perform control on
each phase of the photogrammetric production line in order to
satisfy the requirements set by the consumer of the project.
However, the consumer is interested in final data quality control
in order to infer the real data quality and, consequently, accept
or reject the dataset. The results of data quality evaluation
should become a part of metadata information about the data
set.
The main, very important practical problems to solve in
accomplishing the data quality control are (Kosmatin Fras,
2002):
e selection of proper sampling strategy,
e definition of optimal sample size,
e choice of proper statistical method(s),
e explicit criteria for acception or rejection of the dataset.
4. PHOTOGRAMMETRIC PROCESS
In order to establish a control system for monitoring a
photogrammetric production line, we need to know what
exactly the product should be and parameters affecting the
quality of it, and analyse the production line, i.e. to divide the
process into small phases, find sources of errors and causes of
unsatisfactory performance (Rad, 1995).
The photogrammetric phases and products are graphically
presented in Figure 1. The process usually starts with aerial
survey
camer:
are alr
in the
photos
triangu
points
are ust
equipn
the ne
exterio
data cc
Data «
prepara
proper
Photog
visualiz
product
buildin
Uncerte
the real
and th
propage
or rand
Sources
are oth
reality,