Aiming at the problem of Autofocus window selection in imaging system, a new algorithm based on visual saliency for Autofocus window selection is proposed, which provides a new solution. A re-designed Itti model is used to predict the salient region in the visual scene. By choosing the local maxima in the saliency map to be the seed, the most salient region can be obtained by growing around it and a minimum enclosing rectangle can be found as the focus window. In this paper, the focus window selection based on visual saliency can efficiently capture the visual salient region and the position of general target well, highlight potential focus targets and improves the accuracy of focusing. Compared with the common focusing window selection algorithm, the method proposed in this paper can improve the focusing performance of the imaging system and has wider applicability in the general scene. © 2017 IEEE.