The invention relates to methods and devices for mitigating and/or minimizing effects of hand jitter or shaking during an auto-focusing process of an image-capturing device. In order to more accurately focus a lens for capturing a digital image, the lens is moved between a minimum and maximum position (along an axis) while capturing sample image frames using a focus window at various lens positions. Improved auto-focusing is achieved by dynamically adjusting the relative position of the focus window for each image frame to cover substantially the same image portion of the target image. By adjusting the focus window for each image frame to cover substantially the same image portion of the target image, variations from shaking may be minimized, thereby improving auto-focusing.