Auto white balance based on the similarity of chromaticity histograms
This paper presents a simple and reliable auto white balance algorithm based on the similarity of chromaticity histograms. The proposed algorithm utilizes the property of the r, g and b chromaticity space that only includes the chromaticity information but not the intensity to calculate the chromaticity histograms that describe the chromatic distribution, and compute their similarity measure through Bhattacharyya distances. According to the character that the similarity of chromaticity histograms change following the variation of color temperatures of the light sources, the optimal adjusting coefficients are derived with respect to the maximum of the similarity. Eventually, Through the diagonal transformation matrix composed of the derived adjusting coefficients, the proposed method exactly implements the auto white balance. After experimenting in the real world environment and comparing to other algorithms, the performance of this algorithm is verified and obtains quite improvement as well. Actually the proposed method has been applied in the objects recognition of the outdoor robot in the real-time outdoor environment, and achieved significant effects.
Author's Name: Song, H., Yin, G., Jiang, T.
Volume: Volume 9
Issues: Issue 7
Keywords: Auto white balance, Chromaticity, Color constancy, Histogram, Similarity