经典基追踪模型中所考虑的噪声是加性的高斯白噪声,而实际应用中噪声的形式是多种多样的。因此,经典基追踪模型不能满足处理非高斯噪声环境下的信号去噪问题。基于不同的稀疏性度量函数和不同的拟合误差项形式,对经典基追踪模型进行了扩展,提出了新的基追踪扩展模型,并分析了扩展模型的统计意义。针对其中一类扩展模型,给出了其求解算法。在脉冲噪声环境下的信号去噪实验结果验证了该模型具有比经典基追踪模型更显著的去噪效果。
Traditional basis pursuit model is considered in the presence of additive Gaussian white noise.Eventually,there may be many kinds of noise circumstances in real application.Obviously,traditional basis pursuit model is not enough for non-Gaussian noise circumstances.Based on variable sparseness measure functions and error penalty functions,new extended Basis pursuit model is proposed.Statistical significance of extended basis pursuit model is analyzed,algorithm of a kind of extended basis pursuit model is also brought forward.Some typical signal denosing experiments results demonstrate that the extended Basis Pursuit model can provide better de-noising results than traditional Basis Pursuit model wherein the noise is impulse noise.