Design of a Leaky Integrate-and-Fire Model-Based SPAD Image Sensor for Efficient Edge Detection

Abstract

This paper presents the design of a leaky integrate-and-fire (LIF) model-based single-photon avalanche diode (SPAD) image sensor for efficient edge detection using 110 nm CMOS image sensor technology. To effectively process the digital pulse output of the SPAD, the LIF model was introduced. The LIF model enables real-time analysis through asynchronous event-based information processing, while simultaneously optimizing computational redundancy and power consumption through the utilization of on-chip edge detection processing. The proposed system uses a 16×16 SPAD array and the LIF model for each unit to ultimately output 16×16 resolution 2-D images and 8×8 resolution edge images. To distinguish all edge types (horizontal, vertical, diagonal), 6-type of 2×2 synaptic edge masks with LIF model-based synaptic control scheme are designed. The fabricated chip demonstrated efficient edge detection of 92% accuracy at an operating power as low as 8.1 mW, confirming the validity of the proposed structure.