Wavelet-based focus measure and 3-D autofocusing for microscope images

A new wavelet transform based focus measure was proposed. High numerical aperture (NA) and low NA microscope image sequences were used to compare the performance of this focus measure with the classical and popular focus measures normalized variance, entropy, energy Laplace and other two wavelet transform based high frequency focus measures. The robustness of these focus measures was also compared using image sequence corrupted by Gaussian white noise with that of standard deviation 25. Experimental results show that the proposed focus measure can provide significantly better depth resolution and accuracy than the comparing ones, and exhibit comparatively good robustness with normalized variance. A selective focusing technique for autofocusing on the three dimensional microscale objects in different depth was developed. The results show that this focus measure and selective autofocusing techniques can be widely applied to automated micromanipulation; and that autofocusing is of fundamental importance to automated micromanipulation, and focus measures are the key techniques.