Color Correction by Estimation of Dominant Chromaticity in Multi-Scaled Retinex

r a m n m l t bstract. In image capture a scene with nonuniform illumination as an influence on the image quality, especially the contrast and etail in dark regions. Generally, the tone curve or histogram of an mage is modified to improve the contrast and detail, yet this is inufficient as the intensity and chromaticity of the illumination vary ith geometric position. Thus, the multi-scaled retinex algorithm has een proposed, where the influence of nonuniform illumination is educed by partitioning the original image using local average imges that are estimated based on Gaussian filtering of the original mage. However, the multi-scaled retinex algorithm produces color istortion as the local average images are independently estimated or each channel. In particular, if the chromatic distribution of the riginal image is not uniform and is dominated by a certain chromaicity, the local average image includes not only the intensity and hromaticity of the illumination but also the dominant chromaticity hrough the Gaussian filtering, thereby distorting the color. Accordngly, this article proposes a multi-scaled retinex using a modified ocal average image to reduce the color distortion by the dominant hromaticity of the original image. As with the multi-scaled retinex lgorithm, the local average image is obtained through Gaussian ltering of the original image. The local average image is then diided by the average chromaticity value of the original image to educe the influence of the dominant chromaticity. However, beause the average chromaticity value includes the dominant chroaticity of the original image and the chromaticity of the illumination, he chromaticity removed from the illumination in the local average mage needs to be compensated. Therefore, the chromaticity of the llumination is estimated based on the chromaticity of the highlight egions in the original image. The chromaticity of the local average mage is then modified by the estimated chromaticity. In experients, the proposed method was found to improve local contrast nd reduce the color distortion. © 2009 Society for Imaging Science nd Technology. DOI: 10.2352/J.ImagingSci.Technol.2009.53.5.050502