Chandrasekhar equations and fast algorithms in signal and image processing

摘要

Chandrasekhar equations were introduced in automatic control in the early 70s . Their appearance marked the beginning of audconsiderable amount of research on fast least squares algorithms . They are, however, little used . This is understandable inudautomatic control where speed is not an essential requirement, but is less obvious in signal processing . It may stem from the factudthat their use is linked by an internai approach, or by state space model, to estimation problems which occurs less often inudsignal processing.udThe aim of this article is to take stock of the interest of these equations and the use of the family of algorithms which deriveudfrom them for solving general problems of linear estimation. We shall try to a certain extent to correct the impression givenudabove by showing the links that these equations have with other, more familiar, families of algorithms, and the quasi-canonicudcharacter that they can display in certain applications .