274 |
%depends only on the values of its neighborhood (local characteristics). |
%depends only on the values of its neighborhood (local characteristics). |
275 |
%XXX: resolution-dependency? |
%XXX: resolution-dependency? |
276 |
|
|
|
%% XXX: this is not really texturing: |
|
|
There have been studies on |
|
|
mapping texture appearance to an Euclidian texture space |
|
|
(see \cite{gurnsey01texturespace} and the references therein): |
|
|
in the reported experiments, three dimensions have been sufficient |
|
|
to explain most of the variation in the similarity judgements for |
|
|
artificial textures. |
|
|
However, the texture stimuli have been somewhat simple |
|
|
(no color, lack of frequency-band interaction, etc.). |
|
|
For some natural texture sets, |
|
|
three dimensions have also been |
|
|
sufficient \cite{rao96texturenaming}, but often semantic connections cause the |
|
|
similarity to be context-dependant, making it hard to assess the |
|
|
dimensionality. |
|
|
%% XXX: this is something we should experiment with our textures |
|
277 |
|
|
278 |
% In this article, we apply texture shading to synthesize a large number |
% In this article, we apply texture shading to synthesize a large number |
279 |
% of unique textures for distinguishing virtual objects. |
% of unique textures for distinguishing virtual objects. |
546 |
In these models, a bank of linear filters is applied to the texture followed |
In these models, a bank of linear filters is applied to the texture followed |
547 |
by a nonlinearity and then another set of filters to extract features |
by a nonlinearity and then another set of filters to extract features |
548 |
(see, e.g., Heeger\cite{heeger95pyramid}). |
(see, e.g., Heeger\cite{heeger95pyramid}). |
549 |
There is also physiological evidence of filtering processes: |
There is also physiological evidence of the filtering processes: |
550 |
%The first stages |
%The first stages |
551 |
%of visual perception |
%of visual perception |
552 |
%are fairly well known |
%are fairly well known |
597 |
The simple model we use here assumes |
The simple model we use here assumes |
598 |
that at some point, |
that at some point, |
599 |
the results from the different pre-attentive feature detectors, |
the results from the different pre-attentive feature detectors, |
600 |
such as local and global shapes and colors, |
such as different shapes and colors, |
601 |
are combined to form an abstract \emph{feature vector} |
are combined to form an abstract \emph{feature vector} |
602 |
(see Fig.~\ref{fig-perceptual}). |
(see Fig.~\ref{fig-perceptual}). |
603 |
The feature vector is then used to compute |
The feature vector is then used to compute |
616 |
of noise would all yield |
of noise would all yield |
617 |
{\em almost |
{\em almost |
618 |
exactly the same feature vector} in the brain. |
exactly the same feature vector} in the brain. |
619 |
Noise has no global shapes because there is no correlation between |
Noise has no shape because there is no correlation between |
620 |
the random local features; it is simply perceived as the distribution |
the local features; it is simply perceived as the distribution |
621 |
of the local features, i.e., color and overall frequency |
of texel colors and the overall frequency (the density of texels). |
|
(the density of texels). |
|
622 |
|
|
623 |
From the |
From the |
624 |
model |
model |
637 |
% - seeing different parts of the texture? |
% - seeing different parts of the texture? |
638 |
% - ambiguous perception? |
% - ambiguous perception? |
639 |
\item The entropy of the feature vectors |
\item The entropy of the feature vectors |
640 |
over the distribution of textures, should be maximized. |
over the distribution of textures should be maximized. |
641 |
\end{itemize} |
\end{itemize} |
642 |
|
|
643 |
% which facilitates recognition and memorization of images. |
% which facilitates recognition and memorization of images. |
657 |
%We call this the principle of saving bits. |
%We call this the principle of saving bits. |
658 |
|
|
659 |
The last part means essentially |
The last part means essentially |
660 |
that if all square-like shapes were green, we would again be |
that if all square-like shapes were green, we would be |
661 |
wasting recognitive power. |
wasting recognitive power. |
662 |
There should also be as many possible features in the distribution |
There should also be as many possible features in the distribution |
663 |
as possible. For example, if there were no yellow textures, |
as possible. For example, if there were no yellow textures, |
837 |
them fairly close together around a uniformly chosen mean |
them fairly close together around a uniformly chosen mean |
838 |
but also allows far-away hues |
but also allows far-away hues |
839 |
in the same palette occasionally. |
in the same palette occasionally. |
840 |
The saturations are chosen from distribution emphasizing |
The saturations are chosen from a distribution emphasizing |
841 |
saturated colors: unsaturated colors can easily cause a too multicolored |
saturated colors: unsaturated colors can easily cause a too multicolored |
842 |
palette because the adaptive effects of the eye shift them |
palette because the adaptive effects of the eye shift them |
843 |
towards the complementary colors of the more saturated colors |
towards the complementary colors of the more saturated colors |
844 |
in the palette. |
in the palette. |
845 |
|
|
846 |
XXX: note: maximum saturation of light colors is limited |
XXX: note: maximum saturation of light colors is limited |
847 |
|
|
848 |
% To produce a palette of |
% To produce a palette of |
1396 |
It could also be worthwhile to experiment with other ways |
It could also be worthwhile to experiment with other ways |
1397 |
of visualizing identity, such as unique edge shapes. |
of visualizing identity, such as unique edge shapes. |
1398 |
|
|
1399 |
|
There have been studies on |
1400 |
|
mapping texture appearance to an Euclidian texture space |
1401 |
|
(see \cite{gurnsey01texturespace} and the references therein): |
1402 |
|
in the reported experiments, three dimensions have been sufficient |
1403 |
|
to explain most of the variation in the similarity judgements for |
1404 |
|
artificial textures. |
1405 |
|
However, the texture stimuli have been somewhat simple |
1406 |
|
(no color, lack of frequency-band interaction, etc.). |
1407 |
|
For some natural texture sets, |
1408 |
|
three dimensions have also been |
1409 |
|
sufficient \cite{rao96texturenaming}, but often semantic connections cause the |
1410 |
|
similarity to be context-dependant, making it hard to assess the |
1411 |
|
dimensionality. |
1412 |
|
%% XXX: this is something we should experiment with our textures |
1413 |
|
|
1414 |
% The new graphics chips, ATI R300 and NVIDIA NV30 support |
% The new graphics chips, ATI R300 and NVIDIA NV30 support |
1415 |
% a great deal more of procedural texturing and it will |
% a great deal more of procedural texturing and it will |
1416 |
% be interesting to apply the same criteria there. |
% be interesting to apply the same criteria there. |