Image Processing Reference

In-Depth Information

among
n
possible decisions
d
1
,d
2
,...,d
n
. The main steps we have to achieve in order

to build the fusion process are as follows:

1) modeling: this step includes choosing a formalism and expressions for the ele-

ments of information we wish to fuse within this formalism. This modeling can be

guided by additional information (regarding the information and the context or the

field). Let us assume, to give the reader a better idea, that each source
S
j
provides

an element of information represented by
M
i
regarding the decision
d
i
. The form

of
M
i
depends of course on which formalism was chosen. It can, for example, be a

distribution in a numerical formalism, or a formula in a logical formalism;

2) estimation: most models require an estimation phase (for example, all of the

methods that use distributions). Again, the additional information can come into play;

3) combination: this step involves the choice of an operator, compatible with the

modeling formalism that was chosen, and guided by the additional information;

4) decision: this is the final step of fusion, which allows us to go from information

provided by the sources to the choice of a decision
d
i
.

We will not go into further detail about these steps here because it would require

discussing formalisms and technical aspects. This will be the subject of the following

chapters.

The way these steps are organized defines the fusion system and its architecture.

In the ideal case, the decision is made based on all of the
M
i
, for all of the sources

and all of the decisions. This is referred to as global fusion. In the global model, no

information is overlooked. The complexity of this model and of its implementation

leads to the development of simplified systems, but with more limited performances

[BLO 94].

A second model thus consists of first making local decisions for each source sepa-

rately. In this case, a decision
d
(
j
) is made based on all of the information originating

from the source
S
j
only. This is known as a decentralized decision. Then, in a second

step, these local decisions are fused into a global decision. This model is the obvi-

ous choice when the sources are not available simultaneously. It provides answers

rapidly because procedures are specific to each source, and can easily be adapted to

the addition of new sources. This type of model benefits from the use of techniques

from adaptive control and often uses distributed architectures. It is also referred to

as decision fusion [DAS 96, THO 90]. Its main drawback comes from the fact that

it poorly describes relations between sensors, as well as the possible correlations or

dependences between sources. Furthermore, this model very easily leads to contra-

dictory local decisions (
d
(
j
)

=
k
) and solving these conflicts implies

arbitration on a higher level, which is difficult to optimize, since the original informa-

tion is no longer available. Models of this type are often implemented for real-time

applications, for example in the military.

=
d
(
k
) for
j

Search WWH ::

Custom Search