Neuromorphic circuit and system of bio-inspired vision for the safety enhancement of vehicle applications

This paper describes the bio-inspired neuromorphic vision system, which can mimic the primitive function of visual cortex. The robust performance of proposed neuromorphic system is evaluated by monitoring vehicle passenger’s occupancy, on-vehicle safety apparatus and pedestrians, which is based on the simulated simple cell function of visual cortex experimentation by Hubel and Wiesel as well as the neuromorphic circuit with CMOS linear programmable transconductance. The neuromorphic vision system demonstrated successfully the detection of passenger occupancy and the status of safety seat belt from the images from the widely changing environment of vehicle on the road. The detection of pedestrian is tested with success either for the still image or the video, illustrating the feasibility of this idea to further extended applications utilizing the electronic implementation using CMOS technology. Key-Words: Neuromorphic circuit, Neural networks, Safety belt detection, CMOS, Vision ,Visual cortex, eCall