A robust multi-primitive approach for image mosaicing

Summary form only given. The mosaicing aim is to create relatively to an image set a wide view picture without visible seam-lines. Several algorithms have been proposed to construct mosaics from image sequence where the camera motion is more or less complex. Most of these methods are based on the interest points matching or theoretical corner models. We present a mosaicing method based on regions and Harris point primitives. Indeed, a region matching step is done first in order to limit the search window of the potential-homologous points and so to increase the reliability and robustness of the Harris point matching process by estimating the camera motion (rotation and zooming). Next, resolving a minimization problem, while considering all the matched-point couples, performs the homography. To improve the mosaic continuity around junctions, we apply a radiometric correction. Finally, realized tests on different real sequences show the robustness of the proposed approach.