Stochastic resonance and applications in signal and image processing

Contrary to conventional denoising algorithms, techniques based on stochastic resonance (SR) can improve system performance with the aid of noise. SR techniques have been successfully applied in many areas. We believe that SR can also provide innovation and promising approaches for signal and image processing. It is the focus of this dissertation to investigate the applications of SR-based approaches in these areas.;The SR effect can be realized by either adding noise, or by tuning system parameters (parameter-induced stochastic resonance (PSR)). This dissertation proves, for the first time, that the SR effect can be further enhanced by adding noise and tuning system parameters at the same time for certain nonlinear systems, based on the constrained optimization theory. This will, in turn, further enhance the improvement of system performance.;In order to apply PSR-based bistable detectors (PSRBD) to the tasks with high requirements on both robustness and accuracy, this dissertation investigates the feasibility to use error-correction coding (ECC) techniques together with PSR for this purpose. It is revealed that ECC can only be applied under certain conditions. In addition, the rules used for the design of nonlinear bistable systems and ECC in order to optimize the performance of the whole systems are suggested. This dissertation provides a practical guideline to apply these two techniques together.;This dissertation also proposes an innovative PSR-based signal arrival-time estimator which uses nonlinear bistable systems as pre-processors, and uses both bit-error rate (BER) and waveform pattern matching criteria for the identification of signal locations. It is robust to noise distribution and the length of input signals. In addition, a new nonlinear detector is proposed, which combines the advantages of both matched filter detectors (MFD) and PSRBD.;The theory of two-dimensional parameter-induced stochastic resonance is developed for the first time in this dissertation. A two-dimensional SR system is proposed first, and then the related Fokker-Planck Equation (FPE) and its solutions are derived. The approach based on this theory contributes a different, but promising methodology for nonlinear image processing using SR techniques on the basis of theoretical analysis.