Saul D. Freedman - School of Electrical, Electronic and Computer Engineering The University of Western Australia Perth, Australia
Farid Boussaid - School of Electrical, Electronic and Computer Engineering The University of Western Australia Perth, Australia

In this paper, we propose a CMOS image sensor that exploits multiple capture and spatially varying pixel exposures to achieve a dynamic range of 155 dB, 95 dB higher than a conventional CMOS image sensor. The proposed sensor also utilizes a novel pixel-level counter memory to enhance the sensor's dynamic range and fill factor. Here, the logarithms of counter values used in measurements of incident light intensities are approximated and stored in memory within each pixel. Since these logarithmic values can be represented by a smaller number of bits, each pixel has a smaller area dedicated to memory, allowing a higher fill factor. The proposed CMOS image sensor is well suited to outdoor automotive and mobile robotic applications

CMOS image sensor; fill factor; high dynamic range; logarithmic approximation; multiple capture; spatially varying pixel exposures