A new method, system, apparatus and software product for constructing by an electronic device an image panorama from a plurality of images based on their visual quality is presented. The same portion of a panoramic image scene is typically captured on multiple neighborhood video frames from which one can choose the best visual representation to be pasted into the panorama. Constructing the image panorama from the plurality of captured images can comprise the following steps: a) evaluating image quality of K consecutive frames out of the captured plurality of images using a predetermined criterion, b) selecting one image out of the evaluated K frames with the best image quality using a predetermined rule, and c) appending the selected frame to previously selected frames for constructing the image panorama. On-line and off-line implementations are possible.