Template Adaptation of 2D Quasi-Periodic Data Using a Soft-Assign Localized Correspondence Matrix

In this work we propose a framework for the adaptation of arbitrary quasi-periodic time series. We parameterize and adapt data using traditional free form deformations. The method follows an alternating approach, inspired by the robust point matching algorithm, under which a correspondence matrix is updated and the subsequent deformation is obtained; thus, the template data are adapted to the target data. We demonstrate the performance of the algorithm across several electrocardiogram (ECG) and photoplethysmogram (PPG) databases and compare it to previous works.