About the matlab and C functions 



There are the following files 



G.m: for Gaussian kernel, which is pleasingly smooth

dog.m: for difference of Gaussians center-surround filter, model of dots

DOOG.m: for 2nd derivative difference of offset Gaussians filter, model of bars 

DOOG1.m: for 1st derivative difference of offset Gaussians filter, model of edges

corr.m: for calculating correlation/inhibition coefficients of a filter bank

filters.m: for constructing a filter bank of 136 filters



In matlab, just type 

> filters; 

and see what happen. 



mp.m/Cmp.c: the matlab and C functions for match pursuit



In matlab, compile the Cmp.c by typing 

> mex Cmp.c

Then, read in an image I  (use a small size e.g., 128x128 for speed consideration). 

Now, you can run match pursuit by typing 

> FI = mp(I, F, C, 3.0); 

and see what happens. 



view.m: view the result of match pursuit

synthesize.m/Csynthesize.c: synthesize the image 



In matlab, compile Csynthesize.c by typing 

> mex Csynthesize.c 

Then type 

> v = view(I, FI, F, FS)

and see what happens. 



You can always type >help functionname if you want to know how to use the function

The mp function can be optimized for much faster speed

I have some other codes, but they are less important

