Robin Jenkin received, BSc (Hons) Photographic and Electronic Imaging Science (1995) and his PhD (2001) in the field of image science from University of Westminster. He also holds a M.Res Computer Vision and Image Processing from University College London (1996). Robin is a Fellow of The Royal Photographic Society, UK, and a board member and VP Publications of Society for Imaging Science and Technology. Robin is secretary of the IEEE P2020 Image Quality for Autonomous Vehicles Standards group and leads the sub-group on image quality for viewing. At NVIDIA Corporation, Robin models image quality for autonomous vehicle applications. He is a Visiting Professor at University of Westminster within the Computer Vision and Imaging Technology Research Group.
Title: The Influence of CFA Choice on Automotive and Other Critical Imaging Systems
Abstract: There are a wide variety of color filter array (CFA) colors and combinations available to modern imaging sensors intended for automotive applications. The spectral transmission curve of an individual filter dictates the wavelengths of interest, color saturation and the transparency of the filter and hence is one of the largest influences on the overall sensitivity and color fidelity of the sensor. Recently, combinations of pale filters including clear, yellow, magenta and cyan, have been used with more traditional red, green and blue colors, in an attempt to increase sensitivity. Typically, red, green, blue filters will have total transparency in the 400 to 650nm range of below 35% whereas, clear, yellow, magenta, cyan, in excess of 60%. This increase comes at the expense of effectively reduced color signals leading to larger terms in color correction matrices (CCM) that are subsequently applied and possible cross-correlation of noise between output color channels. Additionally, combining highly transparent channels, such as clear, with those that are less transparent, such as red or blue, can impact performance as exposure has to be reduced to avoid saturation in the most sensitive channel. This can further constrain the less transparent channels.
This work examines this tradeoff and its effect on post-CCM signal quality for a number of CFA combinations and typical automotive spectra of interest, such as traffic lights, signs, and lines. Signal and noise are traced through simulated systems to examine pre- and post-CCM signal quality, via color accuracy, signal-to-noise, modulation transfer function and noise equivalent quanta analysis. Further, the computational cost, benefit, and effect on MTF and post-CCM noise of pre-filtering chroma channels prior to applying the CCM is analyzed. Results are discussed within the context of human viewing and computer vision target applications.