Multiple illuminants with different color temperatures within a scene provide a complicated situation for color constancy and automatic white balance (AWB) algorithms in digital photography. This problem gets even worse in high-dynamic-range (HDR) imaging since a large scale of luminance information is able to be captured and thus it is more likely to be influenced by different illuminants in the scene. Under mixing lighting, a single global adjustment of colors may not yield a good result, since this approach tends to exaggerate the color difference for each illuminant as compared to what observed with the human eye, or only partially remove color cast in the image, making one lighting area look better while others look worse. A local auto white balance algorithm that adjusts colors pixel-by-pixel based on its local area was proposed to solve this problem. For a specific pixel, illumination is estimated from the color information from its neighboring pixels that is weighted by the spatial distance, luminance intensity difference and chromaticity. Experiments on synthetic and real images show that this algorithm performs significantly better than other global and local AWB algorithms when evaluated in terms of the accuracy with which correct surface object colors are estimated.