【求助】用matlab怎么处理超声图象来获得图象分形维数,谢谢!

coffeecat

"图象的分形维数"是什么概念?
回个帖子,如果我能理解,看能不能帮你编一个.

coffeecat

看了一些资料，远不是你给出的公式那么简单。
  
http://www.cindeca.org.ar/programa.htm
  
有工具包，我看可以解决你的问题。你仔细研究研究吧。

coffeecat

另一个参考站点（比较简单）
  
http://www.physics.reading.ac.uk/units/2/2phah/2phah.htm

coffeecat

上面两个地方的程序都可以解决你的问题，
为什么自己不稍稍看一下呢？？？！

lvxiaohong

try this program:
function [H, r, x, y] = hurstVTP (A)
  
   ## usage: [H, r] = hurstVTP (A)
   ##
   ## Computes the Hurst parameter H of the vector A using the variance-time
   ## plot method.
   ##  
   ## Use the variance computed over intervals of different lenghts.  Plotted on
   ## a log-log diagram, var (x) versus the interval length (y) should be
   ## least-squares interpolated by a line of slope beta=2H.  r is the
   ## correlation coefficient and gives a "reliability factor" of the H
   ## estimate: values higher than 0.9 should be expected.
  
   ## I don't know about any data on the precision of this method.  All I know
   ## is that it cannot provide a confidence interval measure.
   ##  
   ## Leland, Taqqu, Willinger, Wilson, "On the Self-Similar Nature of Ethernet
   ## Traffic (Extended Version)", IEEE/ACM Trans. on Networking Vol.2, Num.1,
   ## 1994.
   ##
   ## Put in the public domain by Francesco Potorti` <F.Potorti@cnuce.cnr.it>
   ## Mon Aug 28 17:07:26 MET 1995
  
   if (!is_vector(A))
     error ("A should be a vector");
   endif
  
   # m is the minimum interval length over which the var is computed.  M is
   # the minimum number of intervals over which that same quantity is
   # computed.  k is the number of interval lengths per decade used for the
   # computation.  minp is the minimum number of points that must be used.
   # maxp is the maximum number of points that must be used.
   m = 10;
   M = 10;
   k = 5;
   minp = 5;
   maxp = 20;
  
   r = length (A);
   minr = ceil (m*M*10^((minp-1)/k));
   if (r < minr)
     error (sprintf ("A should have at least %d elements\n", minr));
   endif
  
   # n is the number of points we use to interpolate the line on a log-log
   # plot whose slope is 2H.  x is a vector containing their abscissae, y
   # contains their ordinates, that is, the variance of the increments of the
   # vector A at distances x.
   
   n = min (maxp, floor (k * log10(r/m/M)));
   x = floor (logspace (log10(m), log10(r/M), n))';
   y = zeros (n, 1);
   cumA = cumsum (A);
   for i = 1:n
     xi = x(i);
     Y = cumA (xi:xi:r);
     sets = length (Y);
     Y = Y - prepad (Y(1:sets-1), sets);
     y(i) = (sumsq(Y) - cumA(sets*xi)^2/sets) / (sets-1);
   endfor
   x = log (x);
   y = log (y);
   alfabeta = [ones (size (x)), x] \ y;
   H = alfabeta(2) / 2;
   if (nargout > 1) r = corrcoef (x, y); endif
endfunction
  
--------------------------------------------------------------------------------
有问题请告诉我！