A low-cost VIS-NIR true color night vision video system based on a wide dynamic range CMOS imager
Authors:
- 1 June 2009
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 19310587,p. 273-278
Abstract
Night vision driver assistance systems need to combine high sensitivity with natural color reproduction so that dark scene detail such as pedestrians plus relevant color information from vehicle lights, traffic lights, and retro-reflective signs can be reliably recognized. Dark scene detail is carried by illuminated objects such as pedestrians; color information is carried by colored lights. The intra-scene contrast between dark scene detail and bright colored lights can often be extreme making wide dynamic range (WDR) image capture necessary. Dedicated night vision systems extend the spectral range from the visible (VIS) into the near infrared (NIR) range, but render only a grayscale image, and are not generally WDR-capable. Although multi-spectral VIS-NIR systems can simultaneously capture and fuse color and NIR information, they are often large and costly. A low-cost solution has been developed based on a WDR CMOS imager with widely used color filter array (CFA) technology. In contrast to common practice, no IR cut-filter is used, and the G2-filter of the standard Bayer pattern is removed to create a pixel with VIS+IR sensitivity. This paper concentrates on how to use such an imager in an automotive true color night vision system by removing the IR signal from the color channels then fusing VIS+IR grayscale information with color information.Keywords
This publication has 7 references indexed in Scilit:
- Extended use of ISO 15739 incremental signal-to-noise ratio as reliability criterion for multiple-slope wide dynamic range image capture2009
- A RGBI imager based background suppression method2008
- An adaptive multiple-reset CMOS wide dynamic range imager for automotive vision applications2008
- Colouring the Near-InfraredColor and Imaging Conference, 2008
- OECF characterization of a non-linear HDR Color Camera for Automotive ApplicationsColor and Imaging Conference, 2007
- Quantifying Colour Appearance for Unrelated Colour under Photopic and Mesopic VisionColor and Imaging Conference, 2007
- High dynamic range images from digital cameras raw data2005