A 160×120 Edge Detection Vision Chip for Neuromorphic Systems Using Logarithmic Active Pixel Sensor with Low Power Dissipation

Authors:
Jae-Sung Kong;Dong-Kyu Sung;Hyo-Young Hyun;Jang-Kyoo Shin
Affiliations:
Department of Electronics, Kyungpook National University, Daegu, South Korea 702-701;Department of Electronics, Kyungpook National University, Daegu, South Korea 702-701;Department of Electronics, Kyungpook National University, Daegu, South Korea 702-701;Department of Electronics, Kyungpook National University, Daegu, South Korea 702-701
Venue:
Neural Information Processing
Year:
2008

Citing 4
Cited 0

Analog VLSI and neural systems

Analog VLSI and neural systems
Vision Chips

Vision Chips
Digital Image Processing

Digital Image Processing
An artificial retina chip using switch-selective resistive network for intelligent sensor systems

ICIC'06 Proceedings of the 2006 international conference on Computational Intelligence and Bioinformatics - Volume Part III

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, a vision chip for edge detection based on the structure of a biological retina is introduced. The chip processes an image in a bio-inspired vision mechanism; therefore, it is proper for the neuromorphic systems. Logarithmic active pixel sensor (APS) was applied to the vision chip. By applying a MOS-type photodetector to the logarithmic APS, we could achieve sufficient output swing for the vision chip in natural illumination condition. A correlated-double sampling technique was applied to the chip for noise suppression. In addition, a CMOS buffer circuit is commonly used for both raw and smoothed images by using additional switches. This structure helps reduce the total number of MOSFETs for a unit-pixel and noise. A vision chip with a 160×120 pixel array was fabricated using a 0.35 μm double-poly four-metal CMOS technology, and its operation was successfully demonstrated.