← 返回 JSSC 论文列表
📄 下载 JSSC 原文 PDF
JSSC 2009第9期Image Sensors0.35μm CMOSCMOS Image Sensor

A 200 22s Processing Time Smart Image Sensor for an Eye Tracker Using Pixel-Leve

一种用于眼动追踪的智能图像传感器,采用像素级模拟图像处理技术。
80x60像素阵列,22.4μm像素间距,200μs处理时间,100mW功耗,3.3V供电
CMOS图像传感器眼动追踪浮栅智能传感器像素级处理
创新点1:像素级模拟图像处理技术,通过在像素级别直接进行模拟信号处理,显著减少了传统眼动追踪系统中对复杂外围电路和数字处理器的依赖,从而降低了系统功耗(200μs处理时间)。
创新点2:采用浮栅电容器和非线性电路计算实现像素级处理,这种方法不仅提高了图像处理的精度(±22.4μm误差),还通过模拟计算减少了数字转换的延迟和功耗。
创新点3:低功耗设计,整个系统在3.3V电源电压下仅消耗100mW,适用于移动设备和头戴式显示器(HMD)等对功耗敏感的应用场景。
创新点4:集成化智能CMOS图像传感器(CIS),将图像采集和处理功能集成在同一芯片上,减少了系统复杂度和硅面积(7.5mm²),提高了整体性能和可靠性。
Abstract
, IEEE, and Gunhee Han , Member , IEEE Abstract—This paper presents a smart sensor for an eye tracker using pixel-level analog image processing to eliminate the needs of many peripherals and the high power consumption of the conventional eye trackers. The pixel-level image processing was implemented with floating gate capacitors and nonlinear circuit computations. The prototype chip was fabricated for an 80 60 smart CMOS Image Sensor (CIS) pixel array whose pixel pitch is 22.4 22.4 m/50in a 0.3