Turbo Iterative Signal Processing
Hong Sun, Henri Maı̂tre · 2009
A Turbo iterative method for signal processing is proposed. This method is a kind of multi-systems collaborative signal processing through iteration: several independent systems work in rotation, and each system takes feedback information from the other systems as a priori condition. We have applied such a Turbo iterative signal processing (TISP) method on speech signal enhancement, and on SAR (synthetic aperture radar) image filtering, segmentation and fusion. Some practical results presented in this article show that the Turbo iterative algorithm converges after 5-10 iterations and it improve greatly the signal processing performance. The TISP also shows an effective machine learning method, that is making a discussion between several independent systems through Turbo iteration.