| IP개요 |
This design develops a spiking neural network (SNN) circuit that mimics the human retina’s ability to detect self-motion, object-motion, and looming motion. Retinal ganglion cells (RGCs) process complex visual information, enabling precise motion detection. Traditional vision systems rely on frame-based image recognition, lacking biological efficiency. To address this, we propose a neuromorphic circuit using STDP-driven processing for improved efficiency and lower power consumption. The design follows three phases: (1) modeling retinal motion detection, (2) integrating models into an SNN, and (3) fabricating an energy-efficient CMOS-based neuromorphic circuit. This system enhances motion analysis and has applications in autonomous vehicles, robotics, and surveillance. By unifying self-motion, object-motion, and looming detection, this study advances real-time neuromorphic vision processing, enabling next-generation intelligent systems. |
