Introduction - Hierarchical Neural Networks for Image Interpretation - page 8

structural prototype

vectorization classification

analysis matching

"9"

pixel−image class label

Fig. 1.10. Structural digit classification (image adapted from [21]). Information irrelevant for
classification is discarded in each step while the class information is preserved.

line−drawing structural graph feature−vector

analysis. It consists of a sequence of steps that transform one image representation
into another. Examples for such transformations are edge detection, feature extrac-
tion, segmentation, template matching, and classification. Through these transfor-
mations, the representations become more compact, more abstract, and more sym-
bolic. The individual steps are relatively small, but the nature of the representation
changes completely from one end of the chain, where images are represented as
two-dimensional signals to the other, where symbolic scene descriptions are used.
One example of a bottom-up system for image analysis is the structural digit
recognition system [21], illustrated in Figure 1.10. It transforms the pixel-image of
an isolated handwritten digit into a line-drawing, using a vectorization method. This
discards information about image contrast and the width of the lines. Using struc-
tural analysis, the line-drawing is transformed into an attributed structural graph
that represents the digit using components like curves and loops and their spatial
relations. Small components must be ignored and gaps must be closed in order to
capture the essential structure of a digit. This graph is matched against a database
of structural prototypes. The match selects a specialized classifier. Quantitative at-
tributes of the graph are compiled into a feature vector that is classified by a neural
network. It outputs the class label and a classification confidence. While such a sys-
tem does recognize most digits, it is necessary to reject a small fraction of the digits
to achieve reliable classification.
The top-down approach to image analysis works the opposite direction. It does
not start with the image, but with a database of object models. Hypotheses about the
instantiation of a model are expanded to a less abstract representation by account-
ing, for example, for the object position and pose. The match between an expanded
hypothesis and features extracted from the image is checked in order to verify or re-
ject the hypothesis. If it is rejected, the next hypothesis is generated. This method is
successful if good models of the objects potentially present in the images are avail-
able and verification can be done reliably. Furthermore, one must ensure that the
correct hypothesis is among the first ones that are generated. Top-down techniques
are used for image registration and for tracking of objects in image sequences. In
the latter case, the hypothesis can be generated by predictions which are based on
the analysis results from the preceding frames.
One example of top-down image analysis is the tracking system designed to
localize a mobile robot on a RoboCup soccer field [235], illustrated in Figure 1.11.
A model of the field walls is combined with a hypothesis about the robot position
and mapped to the image obtained from an omnidirectional camera. Perpendicular
to the walls, a transition between the field color (green) and the wall (white) is