Journal: Frontiers in Computational Neuroscience
Article Title: A Biologically Plausible Transform for Visual Recognition that is Invariant to Translation, Scale, and Rotation
doi: 10.3389/fncom.2011.00053
Figure Lengend Snippet: Output of the two-stage transform is sufficient for object recognition . (A) First- and second-stage transforms for various letters. Axes and color code as in Figure E,F. (B) To test recognition, the transform of nine rotated and scaled versions of each letter was compared to the transform of the parent letter. The distances between the 26 parent letters (black letters) and 234 rotated and scaled versions (red dots) are approximately rendered in two dimensions by multidimensional scaling (see ).
Article Snippet: For visualization purposes, all 33,670 of the 10,000-dimensional pairwise distances between the 26 parent and the 234 rotated and scaled letters were plotted in two dimensions using non-classical multidimensional scaling (MDS), as implemented in the Matlab Statistics Toolbox.
Techniques: