A computational model for visual size, location and movement
The ability to detect object size, location and movement is essential for a visual system in either a biological or man made environment. In this paper we present a model for estimating these parameters by using a set of randomly distributed receptive fields on a retina. This approach differs from more conventional ones in which the receptive fields are arranged in a geometric pattern.
The simulation of the model has been performed with a software implementation in a layered fashion. From the input level, computations are performed in parallel which are then combined at a subsequent level to yield estimates of the size and center of gravity of an object. Movement discrimination is implemented by a lateral interaction scheme. The randomly generated receptive fields are now divided into eight weighted classes, corresponding one to a different direction, with the same number of receptive fields for each direction. Both, borders and contrast areas of the object, are useful to identify its motion. When one of the receptive fields detects a border, the weights are changed according to its preferred direction, so that it is possible to follow the movement of the object if it moves this way.
Due to the stochastic nature of the model we can study the effects of receptive field size and density on the results which can be obtained with any desired degree of accuracy. Moreover, since all the parameters are calculated in parallel, based on the same principles and using similar operations, it is possible to have the different parts of the network interact and to make use of results obtained by other subsystems. Finally, in biological systems one also finds some randomness side by side with more deterministic structures. Our model is therefore consistent with this aspect of biological organization.
KeywordsComputational vision visual location motion discrimination
Topics of referenceComputer models of visual systems image analysis neuronal networks
Unable to display preview. Download preview PDF.
- (1).Lettvin J.Y., Maturana H.R., McCulloch W.S. and Pitts W.H.: What the frog's eye tells the frog's brain (1959).Google Scholar
- (3).Leibovic, K.N.: A model for information processing with reference to vision, J. theore. Biology (1966), 11, pp112–130.Google Scholar
- (4).Sobol, I.M.: Lecciones Populares de Matemáticas. Método de Montecarlo. Editorial Nauka-Mir. (1974) [Соболъ, И.М. (1974) Популярные Лекции но МатэмаТИКэ. Метод МонтеКапло. Иэдателщтво Наука.]Google Scholar
- (5).Canavos, G.C.: Probabilidad y Estadistica. McGraw-Hill. (1988)Google Scholar