作者: John A. Perrone
关键词: Optics 、 Optical flow 、 Observer (special relativity) 、 Motion estimation 、 Rotation around a fixed axis 、 Motion field 、 Detector 、 Computer science 、 Computation 、 Machine vision
摘要: I present a method by which direction- and speed-tuned cells, such as those commonly found in the middle temporal area of primate brain, can be used to analyze patterns retinal image motion that are generated during observer movement through environment. For pure translation, is radial nature expands out from point corresponds direction heading. This heading use translation detectors act templates for motion. Each detector sums outputs sensors arranged their preferred lies along center. The most active signifies direction. Rotation constructed similar fashion detect areas uniform speed field produced rotation. A model consisting both types determine independently any rotational observer. achieve this two-dimensional directly does not assume existence accurate estimates It robust aperture problem biologically realistic. basic elements have been shown exist visual cortex.