Location

Gainesville, FL

Start Date

19-5-2023 3:30 PM

End Date

19-5-2023 3:40 PM

Document Type

Extended Abstract

Keywords

Cellular Neural Network, CMOS Image Sensors, Active Pixel Sensor, Analog Preprocessing

Description

A concept for a CMOS image sensor architecture with real-time parallel analog signal processing is presented in this paper. Cellular Neural Networks (CNN) enables image processing in the analog domain using circuit elements such as capacitors, resistors, and nonlinear voltage-controlled current sources. Utilizing CNN for pre-processing in the analog domain allows for real-time feature extraction, noise removal, or other basic image processing, including edge and corner detection. Implementing CNN on a camera requires modification to the CMOS active pixel sensor architecture.

Comments

Revised version based on reviewer's feedback

DOI

https://doi.org/10.5038/IHMA3698

Share

COinS
 
May 19th, 3:30 PM May 19th, 3:40 PM

Implementation of Cellular Neural Network in 5-Transistor Active Pixel Sensors

Gainesville, FL

A concept for a CMOS image sensor architecture with real-time parallel analog signal processing is presented in this paper. Cellular Neural Networks (CNN) enables image processing in the analog domain using circuit elements such as capacitors, resistors, and nonlinear voltage-controlled current sources. Utilizing CNN for pre-processing in the analog domain allows for real-time feature extraction, noise removal, or other basic image processing, including edge and corner detection. Implementing CNN on a camera requires modification to the CMOS active pixel sensor architecture.