Application of a neural architecture to extract motion from image sequences

摘要

Abstract: Investigation of two neural architectures is performedin two dimensions using both synthetic and realimagery. Our model follows the work performed by H.Ogmen and S. Gagne in 1990 on the fly's visual system.We extended their model to a two- dimensionalarchitecture and also developed a new model by addinglong-term memory at the input - adaptive model. Ourinvestigation compares the response of the adaptivemodel against the original Ogmen and Gagne'scell-activity model. The output of both models werefurther processed using casual and noncausal movingaverage filters to help remove tonic image elements andidentify direction of motion. Our simulations show thatthe adaptive model can be used to segment motion fromsequences of imagery.!16