Quantifying focus criterion function comparison for two-stage autofocusing system

Qian Liang, Yufu Qu

Abstract: To obtain high speed and accuracy, passive autofocusing system and two-stage searching strategy have been widely applied in microscopy. The kernel of the passive autofocusing system is the selection of the focus criterion functions, which will significantly affect the efficiency and accuracy of the autofocusing system. In order to help choose the best algorithm, this paper proposes a comparison method for the focus criterion functions in the two searching stages. The ranking methodology for the first stage is proposed for the first time, and the one for the second stage is improved based on previous works. An overall score of quantifying evaluation is also introduced in both stages. The proposed ranking methodology for different focus algorithms is tested on a few synthetic defocused images, which are first simulated from high definition images by Gaussian filter according to the defocus imaging principle, and then added with noise of different ranges. Finally, to verify the effectiveness of the new ranking methodology, real defocused images are captured to evaluate all of these algorithms, the result of which matches the comparison results on synthetic image sets.