Choosing auto focusing evaluation function for images is a key factor for passive auto focusing system of automatic microscope. The basic requirements for a practical auto focusing system are speed

sharpness and robustness to noise. With the relationship between focused and defocused images of a scene

some well known focusing measures (such as gradation variance operator

gradation gradient operator and energy spectrum measure) have been investigated. Based on them

a sum modified Laplacian (SML) operator has been proposed as focusing measures for the first time. The operator is applied to measure the relative sharpness of image sequence at different object distances. Step and threshold are introduced to effectively alleviate the effect of the noise. All of the above mathematical models have been analyzed and compared. Experimental results are presented that demonstrate the accuracy and robustness of the proposed method. The results show that the SML operator is more accurate

stable and reliable than other auto focusing evaluation functions for microscopy images. The algorithm has been applied successfully to automatic focusing system of microscope and testified to be feasible and effective.