This page displays the lattice detection results obtained using the method (and the code) developed by Park et al.:

Deformed Lattice Detection in Real-World Images Using Mean-Shift Belief Propagation
M. Park, K. Brocklehurst, R. Collins and Y. Liu
IEEE Transactions on Pattern Analysis and Machine Intelligence
Special Issue on Probabilistic Graphical Models, 2009

The table below shows example images that we analyzed using the MATLAB code provided by the authorsĀ 
with unmodified parameters. Please note, that we ran the method on all the input images we used in our paper,
so some of them have no visible lattice at all or have other types of symmetry. Of course we do not expect
any lattice detection algorithm to succeed on examples where no lattice exists - these are just included for
completeness.

For each input image in the first column (INPUT) we ran DEMO1.m script, the resulting lattice from the file
*result_result_trial01.jpg is shown in the second column (OUTPUT). Note that for some images the input
result contained no lattice and such images can be found in the bottom. For two input images the script crashed
and did not print the *result_result_trial01.jpg image, in both cases, however, the intermediate iterations printed
in the *_results directory did not contain lattice as well (the images as well as the intermediate iterations can
be found in the table below).

 

INPUT (100%)

OUTPUT (100%)

OUTPUT (75%) OUTPUT (50%)
CRASHED
CRASHED
CRASHED CRASHED

BELOW ARE THE INPUT IMAGES FOR WHICH THE PROGRAM CRASHED
(WE SHOW THE LAST ALGORITHM ITERATION from *_results/result_trial01_it*.jpg
NOTE THAT IN ALL CASES NO LATTICE WAS DETECTED DURING ITERATIONS )

   
   
       

BELOW ARE INPUT IMAGES WHERE NO LATTICE WAS GENERATED
(INPUT IMAGE WAS PRINTED WITHOUT CHANGES FOR ALL SCALES)