Dental implant examination based on the log-polar matching of the maxillary radiograph images in the anisotropic scale space

In this paper a computer vision system is proposed for automatic examination of implant placements based on the maxillary radiograph images. To find rotated and scale changed implants the system does template matching in the extended log-polar space. Matching is proposed to be performed in the anisotropic scale-space, starting from the coarse level. The precise location of an implant is then refined based on the fine level of this space. The two processes are additionally controlled by the contour images which delineate exact positions of implants and other dental works.