The research on automatic white balance for digital microscope

摘要

Automatic white balance plays a key role in digital color imaging system based on CCD/CMOS sensors. The gray world method and its variants are widely used for their simplicity. However, they will fail if the image is dominated by only one or two colors. The iterative method, which extracts gray color points from the image for color temperature estimation, performs well if there are enough gray color points. But it does not work in the case of serious color casts or lack of gray color points. Thus, a new method is proposed combining the iterative method and the gray world method. The iterative method is for fine adjustment, while the gray world method is for coarse adjustment. The characteristics of the digital microscope are taken into account as well. There are three major contributions in the paper. First, brightness constraint is considered during the gray color points detection. The detecting procedure is more precise as a result. Second, each frame of the video stream is divided into n-by-n blocks so as to increase the immunity to the noise. Last, the fine adjustment and the coarse adjustment are combined together. The 'Fine-Coarse-Fine' routine adjusts the image properly even though there are not sufficient gray color points. Experiments on digital microscope indicate that the proposed automatic white balance method is robust, effective and efficient.