next up previous
Next: Gabor model of V1 Up: The Primary Visual Cortex Previous: Spatial Frequency Analysis -

Gabor model of V1 cells I - Spatial Domain

Similar to the Gaussian model for the center-surround receptive field of the retina ganglion cells, here we will develop a more comprehensive model for the V1 cells that can account for multiple aspects of their response properties, such as orientation selectivity, motion (direction and speed) selectivity, and frequency responses in both spatial and temporal domains.

First, we consider the two components in the spatial aspect of the model:

Note that the 2D spatial frequency vector ${\bf\omega}_s=[\omega_x, \omega_y]$ can be expressed by its magnitude $\omega_s$ and direction n (a unit vector along the direction):

\begin{displaymath}\left\{ \begin{array}{l} \omega_s=\sqrt{\omega_x^2+\omega_y^2...
...=[cos \mbox{ } \phi, sin \mbox{ } \phi]
\end{array} \right.
\end{displaymath}

where $ \phi=tan^{-1} (\omega_y/\omega_x)$, and equivalently the two spatial frequency components in $x$ and $y$ can be written as

\begin{displaymath}\left\{ \begin{array}{l} \omega_x=\omega_s \mobx{ } cos \phi \\
\omega_y=\omega_s \mobx{ } sin \phi
\end{array} \right.
\end{displaymath}

The product of the above two functions is a Gabel function which is used to model the receptive field of a V1 cell:

\begin{displaymath}G(x,y)=exp(-\frac{ \vert {\bf r} \vert ^2 }{2\sigma_s^2} ) \mbox{ }
cos[( {\bf r} \cdot {\bf\omega}_s)+\theta]
\end{displaymath}

gabor_sin.gif

gabor_cos.gif


next up previous
Next: Gabor model of V1 Up: The Primary Visual Cortex Previous: Spatial Frequency Analysis -
Ruye Wang 2013-04-08