Night vision systems are becoming more useful as driver assistance systems as the number of accidents during night time is higher than in day time. The images taken in low light conditions are very poor in quality; the pedestrians and animals are hardly visible due to the darkness. In this paper, we have presented two different night vision applications using Generalized Space Frequency Representation (GSFR). The first part of the paper describes a method for image enhancement using GSFR with an exponential kernel. The enhancement includes improvement of brightness, removal of noise and improving the sharpness. The GSFR helps to remove noise from image, brightness and contrast is improved due to the use of the exponential kernel and finally sharpness improvement is done by using unsharp masking based on partial derivative. The second part of the paper describes a method for pedestrian detection using GSFR. It includes calculating weighted average pixel value based on pixel's orientation, applying modified pear shaped curve, applying GSFR, dividing image into layers and then finding a pedestrian in these layers using connected component labelling. © Springer-Verlag Berlin Heidelberg 2011.
CITATION STYLE
Kharade, P., Gindi, S., & Vaidya, V. G. (2011). Generalized space frequency representation techniques for enhancement and detection under low light conditions. In Advanced Microsystems for Automotive Applications 2011: Smart Systems for Electric, Safe and Networked Mobility (pp. 99–107). Kluwer Academic Publishers. https://doi.org/10.1007/978-3-642-21381-6_10
Mendeley helps you to discover research relevant for your work.