Study On Blind Source Separation Techniques And Its Applications

Blind Source Separation (BSS) is the process of recovering unknown independent sources from their linear mixtures without knowing the channel information. The separation is carried out only with the assumption of mutual statistical independence among the source signals. It is widely applied to medical signal processing, speech recognition, image signal processing and communication etc. In recent years, BSS has drawn lots of attention in signal processing community and neural networks community.The performance of separation in strong noisy environment will deteriorate using existed BSS algorithms which work well in the environment without noise. This paper studies the basic model and theory of blind source separation, and proposes a noisy source separation model. The mixed noisy images are separated using the method of wavelet and Curvelet de-noising combined FastICA algorithm. The simulation results show that the method can solve the problem of deterioration of performance of BSS algorithms processing noisy mixtures and well implement the separation of noisy images.Then, in order to solve the problem that the performance of existed BSS algorithm is affected by the non-linear contrast functions, the particle swarm optimization and adaptive particle swarm optimization are introduced. Their characters are good global convergence, fast convergence speed and simple parameters. In this paper, adaptive particle swarm optimization is applied to blind source separation to avoid selecting the non-linear contrast functions. The simulation results show that the proposed method could separate the mixtures of super-Gaussian signals and sub-Gaussian signals.Finally, the model of the signals received by radar array is studied. And the algorithm of blind soure separation based on adaptive particle swarm optimization is applied to separate the signals received by radar array. Computer simulations illustrate the superiority of the algorithm.