Electronically Foveated Dynamic Vision Sensor

This paper proposed a vision system which implements a foveal mechanism to concentrate the attention and dynamically control the center and size of region of interest. The core of the system is an electronically-foveated dynamic vision sensor. An architecture and implementation of an electronically-foveated dynamic vision sensor is proposed. Simulation results demonstrating its operation are provided.

I. Introduction

Conventional vision sensors are based in the acquisition of frames. A conventional pixel in a photograph or video camera integrates the light impinging on the pixel during a subperiod (integration time) of a periodic acquisition time called frame (normally in the order of 20-30ms). Each frame period the sensor will deliver a complete acquired image. When a stimulus moves in the observed scene with a relative speed which is low compared with the periodic frame time, which is normally the case of static background and slow moving parts, the acquired image is very similar from frame to frame containing a lot of redundant information. On the contrary, when a stimulus moves in the observed scene with a relative speed which is high relative to the integration period the object will appear as moved in the acquired image.