Abstract:The current automatic white balance algorithm in which single color modification coefficient is used for all pixels has the deficiency of larger color error. A new automatic white balance algorithm was proposed, which could calculate the color modification coefficient according to the lightness in HSL color space. The HSL color space of an image was analyzed in order to divide the lightness into several segments and pixels in each segment were classified into one group. All pixels near to gray axis in the same group were figured out, the color modification coefficient was calculated based on the average value of color components of all pixels, and then the color of all pixels in this group were modified according to the color modification coefficient. Image data for standard color card in several kinds of color temperature were captured and analyzed respectively according to automatic white balance algorithms, which include HSL gray axis method, gray world method and perfect reflector method. Color error between calculated results of standard cards was analyzed. It showed that HSL gray axis method had much better light environment adaptability and better color reproduction capability.。