This project is an attempt to understand how the visual cortex extracts and represents the structure present in natural scenes. The approach is to formulate visual coding strategies based on theoretical considerations of efficiency and optimality, and to use these codes as the basis for understanding known cortical cell response properties and predicting heretofore unknown properties. There are three parts to this project. The first part will investigate the statistical regularities that occur in natural images and attempt to relate these to the feature selective properties of cortical cells. The second part of the project will be to formulate a neurobiologically plausible model for the development of position- and size-invariant representations of spatial structure. The third part will involve a collaboration with an ongoing neurophysiological investigation in order to formulate and test models of image stabilization (position invariance) in area VI.
| Olshausen, B A; Field, D J (1997) Sparse coding with an overcomplete basis set: a strategy employed by V1? Vision Res 37:3311-25 |
