Classification of the Cardiac Motion using Locally Polynomial Optical Flow Fields computed with the Higher Order Priors

摘要

Modern medical image processing allows us to calculate the motion of the heart. In this paper, we deal with the identification and classification problem of the moving area on the heart. We introduce a method for the classification of the motion using the local order of the flow vectors. For the computation of the polynomially higher order optical flow vectors, we use the higher order prior terms. The orders of spatial differentiation in the prior terms define the polynomial orders local optical flow vectors, that is, the k-th. order prior derives the (k-1) th locally polynomial optical flow field. For example, to compute locally affine optical flow vectors, we are required to adopt the second order prior. The polynomially higher order optical flow fields express the spatially local vibration of the optical flow vectors. Therefore, we classify the motion properties on the heart using this local difference of the vibration in the motion.