Stereo vision is an important feature that enables machine vision systems to perceive their environment in 3D. While machine vision has spawned a variety of software algorithms to solve the stereo-correspondence problem, their implementation and integration in small, fast, and efficient hardware vision systems remains a difficult challenge.
Researchers from the Institute of Neuroinformatics recently presented a new model which solves the stereo correspondence problem robustly exploiting
the temporal dynamics of the event-based vision sensors and provides an efficient solution for machine vision applications. The authors propose a radically novel model that solves the stereo-correspondence with a spiking neural network that can be directly implemented with massively parallel,compact, low-latency and low-power neuromorphic engineering devices.
The model was validated with experimental results, highlighting features that are in agreement with both computational neuroscience stereo vision theories and experimental findings. They demonstrated its features with a prototype neuromorphic hardware system and provided testable predictions on the role of spike-based representations and temporal dynamics in biological stereo vision processing systems. The model also casts new light on the question of how the correspondence problem is solved in the mammalian visual system, and on the role of motion cues in this process.
Successful example of interdisciplinary research
Marc Osswald, Sio-Hoi Ieng, Ryad Benosman & Giacomo Indiveri: A spiking neural network model of 3D perception for event-based neuromorphic stereo vision systems, Scientific Reports 7, Article number: 40703 (2017) , doi:10.1038/srep40703, www.nature.com/articles/srep40703
Prof. Giacomo Indiveri
Institute of Neuroinformatics
Phone: 044 635 30 24