Martin Coath edited Figure_ref_fig_gaborfine_shows__.tex  about 8 years ago

Commit id: 55c554d1b9a95d96682e44be0651080251bf0e5c

deletions | additions      

       

However the \textsc{skv}$_{10}$ in Figure~\ref{fig:gaborfine}(c) produces results which need to be considered carefully. For the 0$^\circ$ and 90$^\circ$ patches the ripples of the Gabor extend far enough in the direction of calculation to appear as large features. As a result, each individual ripple exhibits clear onsets and offsets on this scale, and the \textsc{skv}$_{10}$ result separates in to two dark patches representing the beginning and end of each ripple \textit{and} the patch \textit{as a whole}. The 45$^\circ$ and 135$^\circ$ patches have no features that extend in the direction of calculation so they fade almost to nothing.  The anomalies in the results which favour features that are extended along the horizontal or vertical axis, but not both can, as can (as  was mentioned in the Methods, Methods)  be easily overcome by employing a more sophisticated variant of the method. However the Gabor patches are not typical of features found in natural images and the anomalies observed in other results are much less striking.