A low level real-time vision system using specific computing architectures
Eduardo Ros, Javier Díaz, Suhail M. Odeh, Antonio Cañas · International Conference on Signal Processing · 2006
In this work we present a vision system that includes circuits to compute different modalities in parallel, such as local image features (magnitude, phase and orientation), motion and stereo vision. This becomes possible by efficiently using the massive parallel computing resources of FPGA devices. The paper briefly describes the complete system and discusses the hardware consumption and performance of each visual modality. Finally, the work highlights that huge amount of data produced by such a system and the necessity of on-chip integration mechanisms.