An improved TennenGrad criterion function and an improved Laplace criterion function are proposed for microscopic coarse autofocusing

which use Sobel operators and Laplacian operators to convolve and add together the absolute values of matrix pixels for a image. Several different existed criterion functions in microscope autofocusing are compared and analyzed. To judge the efficiencies of criterion functions

function criteria are analysed with smoothness

efficiency and robustness as the main criteria. In the CCD camera microscope experiment system

a micro accelerometer is used as the measured object. Three autofocusing strategies are proposed and compared

and the

n

-pixel grey level method is preferred for microscopic autofocusing. Coarse autofocusing results show that the improved TennenGrad function has the best performance

autofocusing range is 210 μm

and autofocusing resolution is 7 μm. Image results show that autofocusing curve becomes smoother

and the number of extremums reduce from 5 to 0.