Auto-Focus Algorithm for Moving Target Image

Auto-focus technology is the core part of the automatic optical measure system. Auto-focusing algorithm based on image processing is a long-standing topic in the literatures. The focus-evaluation-function is mathematical description used to measure whether the image is focused or not. The existing sharpness functions will work well for still image sometimes, but in optical tracking measure system, the size of object in image always changing. Thus, these sharpness functions will lose their function. In this paper, an efficient sharpness function based on normalization method of limited gradients is proposed. This auto-focus algorithm calculates the number of gradients bigger than a threshold between the front frame and next frame. This threshold is variable according to image features. According to comparison of the simulating results of several focus-evaluation-functions and combination with practical application, Practical applications illuminate the proposed approach is feasible and effective.