In the example of Section 3.3, we compute that the selectivity is ypt = 2.10~6 if we just use the position of the extremal points and = 1.5.10-8 if we model them using frames. The diameter of the image is d ~ 400 mm and we extracted around 2500 extremal points in each image. We plot in Fig. 9 the number of false positives P with these values.
FIGURE 9 Qualitative estimation of the number of false positives involving at least % matches in MR images of 2500 features. Comparison between frames and points: We need roughly 5 times more point matches than frame matches to obtain the same probability (10 frames and 56 point matches for a probability of 10~10 ).
The actual matches found involve about 500 features and the probability of being a false positive is thus practically zero. However, we must be careful that the "object" we registered is not always the one we wanted, even if this is definitely not a false positive: There can be several different rigid motions in a single image, for instance the skull and the brain in MR images.
Was this article helpful?