IDA
Advanced Signature Analyzer utilizes the
object signature recognition process,
which is comprised of four sequential
steps: 1) identify primitive objects;
2) filter primitives using object rules;
3) merge primitives into “super
objects;” 4) filter “super
objects” using object rules.
The first step in the
process is to identify primitive objects
or clusters based on the proximity of
the defects to neighboring defects and
based on the number of defects in the
cluster. The identification process includes
a line extraction algorithm that will
group the cluster defects that are in
a straight line. An example of this line
extraction function is included in the
sample signature analysis results.
Once the clusters are
identified, the recognition algorithm
creates a shape model for each cluster
based on its defect density distribution.
The algorithm extracts spatial parameters
from the shape to define the primitive
object. Then the recognition function
filters the identified primitive objects
by comparing the object’s spatial
parameters to a set of object rules that
are stored in the object library.
The third step in the
recognition process combines primitive
objects into “super objects”
when they meet specified criteria (for
example, proximity, orientation, X/Y coordinate
value).
The density shape model
is then applied to the “super objects.”
The algorithm extracts another set of
spatial parameters from the “super
object” shape. Then the recognition
function filters the “super objects”
by comparing the spatial parameters to
a set of “super object” rules
that are stored in the object library.
|