|
Next
Shape Understanding
System – visual thinking capabilities of the machine
ZBIGNIEW LES and MAGDALENA LES
The Queen Jadwiga Foundation
This project does not have any financial support from
any Australian Institution. Authors of this project do not receive any payment
for work that is connected with this project.
We are
forced to participate in the ARC Grant “competition” in category a “Private
Researcher”. Until now we have not obtained any clear explanation from ARC
Grant Committee and any other Australian Institution, what does it mean to
be a “private researcher” in Australia.
All
applications for ARC Grant connected with this project were unsuccessful.
It is our filing that the evaluation of our application was not based on
the contents of the proposal. We do not agree with all the procedure in
which applications were processed and evaluated.
ARC
Applications:
2000 Z.
Les. ‘The Method of Shape Understanding– The Study of The Visual Thinking
Possibilities’
2001 L.
Sterling, Z.Les, R. Tadeusiewicz,
L. Latecki, S. Raudys,
B. Falcidieno, M. Pawlak:
‘The Method of Shape Understanding– The Study of The Visual Thinking
Possibilities’
2002 Z.
Les, L. Zadeh, R. Tadeusiewicz,
J. Bezdek: ‘The Method of Shape Understanding–
The Study of The Visual Thinking Possibilities’
This
project is carried out based on the own resources of the authors of the
project; it includes computer equipment, software and all spending
connected with presenting papers at conferences. Outcomes of the research
of this project are donation to The Queen Jadwiga
Foundation by authors of this project.
1. Shape and shape understanding
method
The main novelty of the presented method is that the process of
understanding is related to the visual concept represented as a symbolic
name of the possible classes of shapes. The possible classes of shapes, viewed as hierarchical
structures, are incorporated into the shape model. At each stage of the
reasoning process that led to assigning an examined object to one of the
possible classes, novel processing methods are used. An understanding is
based on interpretation of the visual object as a meaningful unit. A
big advantage of the proposed method of understanding of the visual objects
is that it can explain many problems connected with understanding visual
forms.
The focus of this research is on
shapes that are characteristic of visual objects perceived as
two-dimensional patterns. The shape is considered as a meaningful
unit called a phantom and is related to a meaningful object that in
semiotics is called a sign. The visual objects that are designed to convey
meaning and are used in the computer industry are called icons. All written
languages evolved from pictographic images. Examples of these pictographic
images are hieroglyphs of the ancient Egyptian language. Modern iconic
languages e.g. semantography or ISOTYPE use
visual forms to be able to communicate their meaning.
In shape understanding, shape is the main ‘ingredient’ of the visual
thinking process. However, shape information is complex and its various
components are thoroughly interdependent, and there is no definition of
shape that could include all aspects of shape information. Geometry and
topology, the components of shape information, have
been the subject of study by such mathematicians as Euclid, Pythagoras,
Archimedes, Euler, Mobus, Polya
and Lacatos. The shape not only determines how an
object looks, but also forms the basis for many of its other properties.
Webster defines shape as ‘that quality of an object which depends on the
relative position of all points composing its outline or external surface.’
The shape is also described as a silhouette of the object (e.g. obtained by
illuminating the object by an infinitely distant light source). Marr treats shape as one of the forms
of an object representation or a ‘special visual’ feature of an object.
Perception research lays emphasis on the use of boundaries in decomposing
objects into their parts, especially on describing rules for detection of
part boundaries, e.g. based on notions as ‘concavities’ of concave regions
or the ‘minimal rule’. Based on the geometrical permissions Leyton developed a theory that claims that all
shapes are basically circles, which changed form as a result of various
deformations caused by external forces like pushing.
In this research the shape is considered as a meaningful unit called
a phantom and is related to a meaningful object that in semiotics is called
a sign. A shape understanding system is designed to become an independent
module that performs a shape understanding task. Understanding involves a
transformation of the data given in the form of critical points into a set
of descriptors and next into the sub-symbolic and symbolic representations.
The shape understanding system (SUS) consists of two main modules: the
central module (the master expert and the reasoning expert) and the
peripheral module (the generating expert, the query expert, the
self-correcting expert, the learning expert and the spatial-logic expert).
2. Visual perception
Shape understanding is an activity that makes available a number of
different visual, semantics and verbal descriptions. Normally, when we
recognize a particular instance of a shape category, we are able to
describe how the object would look if seen from a different viewpoint. Shape understanding method is a
multidisciplinary research area that is focused on the understanding of the
visual objects. Understanding of the visual object is related to research
in the area of human visual perception
3. Understanding the task
Understanding is power that enables objects to be thought.
Understanding and thought were topics of philosophical thinkers (e.g. Locke,
Hume, Berkeley, Leibnitz, Kant, Popper). The
classical preoccupation with forms and universals was concerned with
generality of thought. This spawned a whole class of general or potentially
general entities such as forms, universals, essences and sensible species.
Thoughts and their contents possess two strictly distinct but closely
connected properties: intentionality - they are about things other than
themselves and generality - nearly all our concepts express features which
an indefinite number of things might posses.
According to Aquinas, the direct object of human intellectual
knowledge is the form abstracted from matter, which is the principle of
individuation, and known through the universal concept. Scotus
discarded the traditional Augustinian-Franciscan theory of a special divine
illumination and held, with Aquinas, that Aristotelian doctrine of the
abstraction of the universal can explain the genesis of human knowledge
without it being necessary to invoke either innate ideas or a special
divine illumination. According to Kant understanding is one of the higher faculties of knowledge.
In hermeneutics understanding is the inversion of a speech act,
during which the thought that was the basis of the speech must become
conscious. More recently, Jean Piaget specified four stages through which
he said individuals construct an understanding of reality by means of
internalized, reversible mental operations that act on the world to produce
a cognitive independence from physical appearance.
Visual Understanding involves:
1.
Knowing what is meant or intended by a visual
object. In the case of a perception of the real world object, it can be an
obstacle on the road or a block to be moved, whereas in the case of an icon
it can be a description of the scene.
2.
Knowing how it works. It involves schematic drawing
(engineering, anatomy).
4. The shape understanding system
The proposed shape understanding system operates on the knowledge of image
processing, decision making and search strategies as well as the knowledge
of shape description and representation distributed among the specialized
experts. SUS consists of the two main modules: the central reasoning module
and the peripheral module. The central reasoning module consists of the
master expert, the reasoning expert, the manager expert, the processing
expert and the end-the analysis expert. The peripheral module consists of
the generating expert, the question expert, the self-correcting expert, the
learning.
Next
|
