Abstract:Traditional computer-vision technology that relies on image sensors coupled with cloud processing or on-chip Artificial Intelligence (AI) processors have encountered significant challenges in terms of power consumption, delays arising from data transmission, and/or memory access. In-sensor and near-sensor computing have been reported to solve this issue by applying pixel or array level feature extraction [2–6]. In [2, 4], capacitors are utilized for analog-domain Haar filtering to reduce the processing power consumption, but sacrificing the Fill Factor (FF) due to the use of the capacitor array [2], or complicated pixel-level logic [4]. An image sensor in introduced in [3] with Hog feature-based object detection, achieving both low power and high accuracy for the detection of up to three object classes. However, it does not work on more complex tasks. Convolutional Neural Networks (CNNs) [5, 6] offer enhanced accuracy in practical applications, but increase the power consumption and circuit complexity. Optical-domain processing [7] is capable of parallel information processing through an optical architecture based on free space. However, in [7] this is based on diffraction and requires coherent light as input, which constrains the practicality for edge computing. This paper presents a Pulse Width Modulation (PWM) pixel-based image sensor array that integrates an optical-electronic hybrid 3-layer convolutional processing unit. The first layer of the convolution is performed in the optical domain without introducing extra power consumption. The integration of optical convolution with electrical convolution offers the potential to leverage the energy-efficient advantages of optical convolution, while simultaneously providing the essential programmability needed to address multiple use-cases through reprogramming within a single device. A dual-mode Processing Element (PE) is proposed enables both acquisition mode and calculation mode. The integration of a sum/differential dual-counter design and the incorporation of two different pixel reset methods are employed in the co-design of PWM pixels and PEs to effectively mitigate computational power consumption.

A Single-Slope PWM Imaging Method for Multi-Mode Dynamic Vision Sensor

A Reconfigurable Convolution-in-Pixel CMOS Image Sensor Architecture

A Cmos Image Sensor For Monochromatic Spectrum Imaging

A Multi-mode Neuromorphic Vision Sensor With Improved Brightness Measurement Performance by Pulse Coding Method

A Multimode Neuromorphic Vision Sensor with Improved Brightness Measurement Performance by Pulse Coding Method

A 128$\,\times$ 128 1.5% Contrast Sensitivity 0.9% FPN 3 µs Latency 4 mW Asynchronous Frame-Free Dynamic Vision Sensor Using Transimpedance Preamplifiers

A Composite Photodector with Wide Dynamic Range and Small Area for Dynamic Vision Sensor Application

Single-pixel Multispectral Imaging Based on Macropixel Segmentation Method

An improved pixel circuit with low noise event rate and enhanced bright‐light sensitivity for dynamic vision sensor

A dual threshold adaptation method to dynamic vision sensor pixel

An Over 140 dB Dynamic Range CMOS Image Sensor Combined DCG and Logarithmic Response

320×240 Pixels CMOS Digital Image Sensor with Wide Dynamic Range

A 128x1 Pixels, High Dynamic Range Spad Imager In 0.18 Mu M Cmos Technology

A <tex>$128\times 1$</tex> Pixels, High Dynamic Range SPAD Imager in <tex>$0.18\ \mu\text{m}$</tex> CMOS Technology

An Intrascene Wide Dynamic Range CMOS Image Sensor Using Dual-Gain Ramp Generator for Machine Vision Applications

A $128\times 1$ Pixels, High Dynamic Range SPAD Imager in $0.18\ \mu\text{m}$ CMOS Technology.

Novel CMOS image sensor with a 132-dB dynamic range

Circuit Design of High-Dynamic-Range Image Sensor Pixel for Visual Prostheses

6.9 A 0.35V 0.367TOPS/W Image Sensor with 3-Layer Optical-Electronic Hybrid Convolutional Neural Network

Novel Logarithmic Active Pixel Sensor with High Dynamic Range and High Output Swing

A Dual-Mode CMOS Image Sensor Based on In-Pixel Frame Differencing