IMAGE REGISTRATION BASED ON KEY-POINT MATCHING

摘要

There is provided mechanisms for image registration. A method is performed by an image registration entity (100). The method comprises obtaining (S102) a matching between a first set of objects (320P, 330P, 340P, 350P) in a first image (310P) of a scene and a second set of objects (320C, 330C, 340C, 350C) in a second image (310C) of the scene. The first set of objects and the second set of objects are extracted using visual object detection in the first image and in the second image, respectively. The method comprises obtaining (S104) a first set of key-points as extracted from the first image (310P) and a second set of key-points as extracted from the second image (310C). The method comprises performing (S106) image registration by matching the first set of key-points to the second set of key-points. Those of the first set of key-points that are mapped to objects in the first set of objects and that have matching objects in the second set of objects are restricted to only be matched to those of the second set of key-points that are mapped to any of the matching objects in the second set of objects. The method comprises applying (S108) the matching between key-points resulting from the image registration when constructing an image representation of the scene.