Title :
A Bio-Inspired APS for Selective Visual Attention
Author :
Cottini, Nicola ; Gottardi, Massimo ; Massari, Nicola ; Passerone, Roberto
Author_Institution :
Fondazione Bruno Kessler, Trento, Italy
Abstract :
This letter presents an active pixel sensor architecture that mimics the eye adaptive behavior through dynamic background subtraction as a basis for image interpretation tasks. The pixel extracts temporal contrast as a low-level visual feature for successive high-level processing targeted to monitoring applications. To do this, each pixel compares the acquired value with a voltage range computed over the past pixel behavior; values outside this range signal a potential alert. Two switched-capacitor programmable low-pass filters are embedded. The 26 μm, 12% fill-factor, fully functional 45 T pixel has been fabricated in a 0.35 μm CMOS technology.
Keywords :
CMOS integrated circuits; biosensors; cameras; feature extraction; image sensors; low-pass filters; switched capacitor filters; CMOS technology; active pixel sensor architecture; bioinspired APS; dynamic background subtraction; image interpretation tasks; low-level visual feature; pixel compares; selective visual attention; size 0.35 mum; successive high-level processing; switched-capacitor programmable low-pass filters; temporal contrast extraction; Vision sensors; active pixel sensors; background subtraction; early visual features; visual attention;
Journal_Title :
Sensors Journal, IEEE
DOI :
10.1109/JSEN.2013.2263962