It ' s a problem of great significance how to filtering a signal in joint time - frequency domain and reconstruct the time series from the time - frequency plane selected . in this thesis , the theory of joint time - frequency analysis and synthesis and time - frequency filtering is investigated . two kinds of new time - frequency domain filtering methods are proposed based on adaptive decomposition and linear subspace projection 本文在系统学习现有的各种时频分析及其综合算法的基础上,详细研究了时频域滤波的核心思想和基本方法,吸取了自适应时频分析和线性子空间投影wigner - ville变换的优点,提出了的两种新的时频域滤波方法,并通过理论分析和数字仿真完成了相应的数值特性与应用特性研究。