Adaptive resolution image acquisition using image mosaicing technique from video sequence

We propose a layered image mosaicing method from a video sequence to acquire an adaptive resolution image. In research of art, for example, researchers often desires the full view of a painting and the details of certain regions, where fine brush-strokes are in it. The proposed method is to acquire the region of special interest at a fine resolution and the remaining image space at a coarser resolution by using a hand-held video camera which repeats moving, panning, tilting, zooming in and zooming out. To integrate video frames, image mosaicing methods can be used. However, existing mosaicing methods, in general, have a defect, where they perform inefficiently because wide angle image frames are registered after magnification. To solve this problem, the layered image mosaicing method is proposed. The proposed method can reduce processing time while it makes a mosaic image accurately. In this paper, the concept, algorithm and experiments are described. The experiments showed that the method was efficient and accurate for acquisition of an adaptive resolution image.