作者: Alexis Quesada-Arencibia , Roberto Moreno-Díaz , Miguel Aleman-Flores
关键词: Receptive field 、 Object (computer science) 、 Machine vision 、 Motion estimation 、 Computer science 、 Retina 、 Weighting 、 Computer vision 、 Motion analysis 、 Artificial intelligence
摘要: In this paper we present a working vision system for estimating size, location and motion of an object by using set randomly distributed receptive fields on retina. The approach used here differs from more conventional ones in which the are arranged geometric pattern. From input level, computations performed parallel two different channels: one purely spatial properties, other time-space analysis, then at subsequent level to yield estimates size center gravity speed direction motion. Movement analysis refining is implemented lateral interaction (spatial) memory (temporal) schemes build trajectory. parameters involved (receptive field weighting function, number cells) tested speeds results compared, yielding new insights functioning living retina suggesting ideas improving artificial system.