Abstract
In these days, vehicle providers are developing many convenience applications for driver. These applications are required reliability regardless of whether condition, such as heavy rain, deep fog, and so on. Light source blurring is one of these weather condition problems. In this paper, we presented the new light source blurring correction method in night foggy vehicle image. Night fog weather condition make light blur effect due to high dynamic range characteristic and scattering property. We preferentially defined light blur image in vehicle image. Light blur vehicle image have characteristics that are high dynamic range property, colored light blur region, low contrast due to fog. To compensate these characteristics, we divided the region into light blur segment using its channel property. For each segment, we find the light source invariant image by projection to its opposite light source color, and extract edge that is hided by light source blur. We aimed at improving edge extraction property because the vision application that is used by front vehicle camera is required accurate edge. To analysis our algorithm, we compared the result of our algorithm result with the result of previous method result. And we show proposed algorithm superior to previous algorithms through various experiments.
Keywords: Intelligent vehicle, Vehicle image, Weather condition