The problem of blind separation of signals in post nonlinear mixture is addressed in this paper. The post nonlinear mixture is formed by a component wise nonlinear distortion after the linear mixture. Hence a nonlin...The problem of blind separation of signals in post nonlinear mixture is addressed in this paper. The post nonlinear mixture is formed by a component wise nonlinear distortion after the linear mixture. Hence a nonlinear adjusting part placed in front of the linear separation structure is needed to compensate for the distortion in separating such signals. The learning rules for the post nonlinear separation structure are derived by a maximum likelihood approach. An algorithm for blind separation of post nonlinearly mixed sub and super Gaussian signals is proposed based on some previous work. Multilayer perceptrons are used in this algorithm to model the nonlinear part of the separation structure. The algorithm switches between sub and super Gaussian probability models during learning according to a stability condition and operates in a block adaptive manner. The effectiveness of the algorithm is verified by experiments on simulated and real world signals.展开更多
Nonlinear spectral mixture analysis (NSMA) is a widely used unmixing algorithm. It can fit the mixed spectra adequately, but collinearity effect among true and virtual endmembers will decrease the retrieval accuracies...Nonlinear spectral mixture analysis (NSMA) is a widely used unmixing algorithm. It can fit the mixed spectra adequately, but collinearity effect among true and virtual endmembers will decrease the retrieval accuracies of endmember fractions. Use of linear spectral mixture analysis (LSMA) can effectively reduce the degree of collinearity in the NSMA. However, the inadequate modeling of mixed spectra in the LSMA will also yield retrieval errors, especially for the cases where the multiple scattering is not ignorable. In this study, a generalized spectral unmixing scheme based on a spectral shape measure, i.e. spectral information divergence (SID), was applied to overcome the limitations of the conventional NSMA and LSMA. Two simulation experiments were undertaken to test the performances of the SID, LSMA and NSMA in the mixture cases of treesoil, tree-concrete and tree-grass. Results demonstrated that the SID yielded higher accuracies than the LSMA for almost all the mixture cases in this study. On the other hand, performances of the SID method were comparable with the NSMA for the tree-soil and tree-grass mixture cases, but significantly better than the NSMA for the tree-concrete mixture case. All the results indicate that the SID method is fairly effective to circumvent collinearity effect within the NSMA, and compensate the inadequate modeling of mixed spectra within the LSMA.展开更多
In this paper, an evolutionary recursive Bayesian estimation algorithm is presented, which incorporates the latest observation with a new proposal distribution, and the posterior state density is represented by a Gaus...In this paper, an evolutionary recursive Bayesian estimation algorithm is presented, which incorporates the latest observation with a new proposal distribution, and the posterior state density is represented by a Gaussian mixture model that is recovered from the weighted particle set of the measurement update step by means of a weighted expectation-maximization algorithm. This step replaces the resampling stage needed by most particle filters and relieves the effect caused by sample impoverishment. A nonlinear tracking problem shows that this new approach outperforms other related particle filters.展开更多
文摘The problem of blind separation of signals in post nonlinear mixture is addressed in this paper. The post nonlinear mixture is formed by a component wise nonlinear distortion after the linear mixture. Hence a nonlinear adjusting part placed in front of the linear separation structure is needed to compensate for the distortion in separating such signals. The learning rules for the post nonlinear separation structure are derived by a maximum likelihood approach. An algorithm for blind separation of post nonlinearly mixed sub and super Gaussian signals is proposed based on some previous work. Multilayer perceptrons are used in this algorithm to model the nonlinear part of the separation structure. The algorithm switches between sub and super Gaussian probability models during learning according to a stability condition and operates in a block adaptive manner. The effectiveness of the algorithm is verified by experiments on simulated and real world signals.
文摘Nonlinear spectral mixture analysis (NSMA) is a widely used unmixing algorithm. It can fit the mixed spectra adequately, but collinearity effect among true and virtual endmembers will decrease the retrieval accuracies of endmember fractions. Use of linear spectral mixture analysis (LSMA) can effectively reduce the degree of collinearity in the NSMA. However, the inadequate modeling of mixed spectra in the LSMA will also yield retrieval errors, especially for the cases where the multiple scattering is not ignorable. In this study, a generalized spectral unmixing scheme based on a spectral shape measure, i.e. spectral information divergence (SID), was applied to overcome the limitations of the conventional NSMA and LSMA. Two simulation experiments were undertaken to test the performances of the SID, LSMA and NSMA in the mixture cases of treesoil, tree-concrete and tree-grass. Results demonstrated that the SID yielded higher accuracies than the LSMA for almost all the mixture cases in this study. On the other hand, performances of the SID method were comparable with the NSMA for the tree-soil and tree-grass mixture cases, but significantly better than the NSMA for the tree-concrete mixture case. All the results indicate that the SID method is fairly effective to circumvent collinearity effect within the NSMA, and compensate the inadequate modeling of mixed spectra within the LSMA.
基金Sponsored by the National Security Major Basic Research Project of China(Grant No.973 -61334)
文摘In this paper, an evolutionary recursive Bayesian estimation algorithm is presented, which incorporates the latest observation with a new proposal distribution, and the posterior state density is represented by a Gaussian mixture model that is recovered from the weighted particle set of the measurement update step by means of a weighted expectation-maximization algorithm. This step replaces the resampling stage needed by most particle filters and relieves the effect caused by sample impoverishment. A nonlinear tracking problem shows that this new approach outperforms other related particle filters.
