Reconstruction from missing function and derivative samples and oversampled filter banks

The paper deals with two-channel sampling using an oversampled filter bank. It emphasizes sampling of a function, f, and its derivative, f´, at a rate higher than the critical minimum rate (oversampling), and the problem of reconstructing the function even when a finite number of samples (of f or f´) are unknown. The motivation for considering this problem is the need to add redundancy to data, as in conventional channel coding, and the convenience of doing it in a domain suited to the data, as in joint source-channel coding.