Smart pixel technology and an application to two-dimensional analog-to-digital conversion

A short review of smart pixel technology is followed by an application of o ne specific smart pixel technology to a 2-D application of analog-to-digita l conversion called digital image halftoning. A novel approach to digital i mage halftoning is described based on a symmetric error diffusion algorithm , a new form of artificial neural network called an error diffusion neural network, and a smart pixel optoelectronic architecture. Two generations of smart pixel architectures are described that incorporate self-electro-optic effect device (SEED) modulators flip-chip bonded to submicrometer feature size complementary metal-oxide semiconductor (CMOS) silicon circuitry to pr oduce CMOS-SEED implementations of the error diffusion neural network. Simu lations and experimental characterizations demonstrate that this hardware a pproach provides sufficient computational accuracy for the analog neural ne twork while simultaneously providing switching speeds that support halftoni ng at video rates, (C) 1998 Society of Photo-Optical Instrumentation Engine ers. [S0091-3286(98)00112-3].