/[gzz]/manuscripts/Paper/paper.tex
ViewVC logotype

Diff of /manuscripts/Paper/paper.tex

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 1.14 by jvk, Sat Mar 8 10:12:22 2003 UTC revision 1.15 by jvk, Sat Mar 8 10:27:59 2003 UTC
# Line 256  proposals of texture discrimation models Line 256  proposals of texture discrimation models
256  $N$th order statistics of pixels and elaborate connectivity  $N$th order statistics of pixels and elaborate connectivity
257  structures of certain micropatterns.XXX  structures of certain micropatterns.XXX
258    
259  Textons \cite{julesz81textons}: elongated blobs, line terminators,  Statistical modeling of textures as samples from a probability
260  line crossings, etc.  distribution on a random field as already seen in \cite{julesz62visualpattern}
261    in a very simple form.
262    The most popualar computational approach is Markov random fields
263    \cite{cross83markov, geman84stochastic}, where a texture
264    is characterized by its local statistics.
265    However, these approaches are often not feasible on large
266    large neighborhoods, but essentially work on pixel scale.
267    
268    Attempt to explain texture perception by the densities of textons
269    \cite{julesz81textons}, fundamental texture elements, such as
270    elongated blobs, line terminators, line crossings, etc.  
271    However, the textons are hard to define formally.
272    
273    Textures are continuous
274    
275  Of course, such models are not directly applicable on high-resolution  Of course, such models are not directly applicable on high-resolution
276  textures; some kind of filtering would be required to obtain the input.    textures; some kind of filtering would be required to obtain the input.  
# Line 265  However, the filtering itself can also h Line 278  However, the filtering itself can also h
278  \cite{bergen88earlyvision}.XXX  \cite{bergen88earlyvision}.XXX
279    
280  Filtering based approach, e.g., \cite{heeger95pyramid}.  Filtering based approach, e.g., \cite{heeger95pyramid}.
281    Essentially a bank of linear filters is applied to the texture followed
282    by a nonlineary and then another set of filters.
283    
284    XXX: reviews
 Statistical modeling of textures as samples from a probability  
 distribution on a random field as already seen in \cite{julesz62visualpattern}  
 in a very simple form.  
 The most popualar approach is Markov random fields  
 \cite{cross83markov, geman84stochastic}, where a texture  
 is characterized by its local statistics.  
285    
286  \subsection{Focus+Context views}  \subsection{Focus+Context views}
287    

Legend:
Removed from v.1.14  
changed lines
  Added in v.1.15

savannah-hackers-public@gnu.org
ViewVC Help
Powered by ViewVC 1.1.26