A Computer Simulated Model of Visual Processing.
NAVAL UNDERWATER SYSTEMS CENTER NEWPORT R I
Pagination or Media Count:
The work performed concerns structural models since it porposes a neutral structure to account for a wide range of behavior and shows that the psychophysical and physiological data are compatible with the proposed structure. The model casts some double on the validity of feature extractors which have assumed a high degree of specificity in cortical neurons, based on physiological data alone. In precortical layers the neural network is a logical extension of the Hartline-Ratliff model. A convergent overlapping structure of fiber bundles is proposed in the cortex. A theoretical discussion of the anticipated behavior is presented and two mathematical constraints on the model are discussed. The practical aspects of simulating the model are reviewed and the modeling techniques are described. The model was simulated on a general purpose computer using the FORTRAN V language. The experimental results show that the model exhibits the behavior of brightness compression, contour enhancement, circular receptive fields in precortical layers and straight-edged receptive fields in the cortex. Author