Abstract: The design and selection of focusing evaluation function was one of the key problems of automatic focusing microscopes. The illumination change would cause function curve to lose its ideal characteristics. The traditional focusing evaluation function can not overcome this problem. In order to solve the problem, an automatic focusing algorithm combining discrete cosine transform (DCT) coefficient of zero of frequency domain and local standard deviation of spatial domain was proposed. DCT coefficient of zero and local standard deviation of sub-blocks were calculated. Division operation was performed by taking advantage of the opposite effect. And square algorithm was used to increase the clarity ratio of the algorithm. Under different light conditions, compared with several traditional focusing evaluation functions, the focusing experiment was carried out to verify the applicability of the new algorithm. The performance of focus evaluation function was evaluated quantitatively. The results show that the algorithm can keep good curve characteristics under low illumination conditions and has strong anti-noise ability, sensitivity and stability.