Abstract: The common automatic white balance algorithm based on histogram is failure when there are great different between there color histograms. An automatic white balance based on histogram matching is applied. The histogram matching is used when three channels have high similarity in histogram. While the histogram similarity is low, histogram translation method is used to correct images respectively. The combination of histogram matching and histogram translation algorithm makes up for the deficiency of the two algorithms effectively. Experimental results show that, on a large number of natural images set, the method has better color balance effect and the applicable range of the algorithm is increased by 11%.