Region Covariance for Deformation Measurement of Non-rigid Biological Tissues from MR Images

In this paper, we propose an integrated method SRC (SIFT and Region Covariance) for feature point matching in MR images of nonlinear and nonuniform deformation of nonrigid tissues. First, from the template and deformed images, the DoG detector and SIFT descriptor are used to extract and describe the feature points. Then use the covariance of neighborhood feature statistics computed inside a region of interest as the region descriptor. Finally, use a nearest neighbor search algorithm using a distance metric defined on the positive definite symmetric matrices for feature matching. The experimental results showed that the proposed SRC method outperformed the single SURF and SIFT methods.