Construction and evaluation of a wavelet-based focus measure for microscopy imaging.

Microscopy imaging can not achieve both high resolution and wide image space simultaneously. Autofocusing is of fundamental importance to automated micromanipulation. This article proposes a new wavelet-based focus measure, which is defined as a ratio of high frequency coefficients and low frequency coefficients. 8 series of 49 microscope images each acquired under five magnifications are used to comprehensively compare the performance of our focus measure with the classic and popular focus measures, including Normalized Variance, Entropy, Energy Laplace and wavelet-based high frequency focus measures. The robustness of these focus measures is evaluated using noisy image sequences corrupted by Gaussian white noise with standard deviations (STD) 5 and 15. An evaluation methodology is proposed, based on which these 5 focus measures are ranked. Experimental results show that the proposed focus measure can provide significantly the best overall performance and robustness. This focus measure can be widely applied to the automated biological and biomedical applications.