“Computation and correction of dynamic white balance in the automotive cameras”

Dynamic White Balance (DWB) correction is essential for detection and recognition of color-based entities such as colored lanes in automotive camera. This kind of correction addresses problems like color deviation due to effect of un-intended illumination. Automotive camera undergoes a lot of real time situation deals with one or more illumination sources present in the scene apart from background illumination. Moreover, reference white balance measurement targets may contain different artifact. This results the color reproduction capability of automotive imaging system to generate the deviated color space than expected and changes dynamically. As stated above, measurement and correction of such color space deviation (dynamic white balance) is an important requirement in automotive cameras. This paper highlights a novel method derived with backward frame difference model algorithm applied on segmented road region of interest (ROI) of road. The said method implements a color histogram image processing algorithmic logic to obtain more accurate dynamic while balance for automotive camera. This method also addresses the DWB computation / correction independent of target artifacts and effect of strong illuminating sources, in most of real-world automotive scenarios. The literature survey, problem definition, theoretical investigation and experimental results are presented in this paper. There is approximately ~18% improvement in computation of white balance ratios than traditional gray world method in the max-min color difference image.