Matlab7.x

用Matlab中的Haar和Db9小波对图像进行分解和重构,并在不同阈值下用PNG格式存储重构的图像.(Matlab Haar and the right image Db9 wavelet decomposition and reconstruction, and under different thresholds with PNG format image storage remodeling.)

This gives a brief discussion on speech processing

Kalman滤波程序,在一维的数据上，观看滤波后估计值与预测值之间的差距。(Kalman filtering)

Transcoder implementation

DTW算法的程序，申请两个n*m的矩阵D、d,分别为累积距离和帧匹配距离。这里n和m为测试模版与参考模版的帧数。然后通过一个循环计算两个模版的帧匹配距离d。接下来进行动态规划，为每个格点 （i，j）都计算其三个可能的前续格点的累积距离D1，D2，D3。考虑到边界问题，有些前续格点可能不存在，因此加入一些判断条件最后利用最小值函数min()，找到三个前续格点的累积 距离作为累积距离，与当前帧的匹配距离d（i，j）相加，作为当前格点的累积距离。该计算过程一直达到格点（n，m），并将D（n，m）输出，作为模版匹配的结果。(DTW algorithm procedures, apply two n* m matrix D, d, respectively, accumulated distance and frame matching distance. Where n and m is a reference test pattern and the template frames. Then through a loop to calculate two frames template matching distance d. Next, dynamic programming, for each grid point (i, j) are calculated from the cumulative D1 before the renewal of its three possible lattice point, D2, D3. Taking into account the boundary problem, some former continued grid point may not exist, so add some conditions to determine the final use of the minimum function min (), found three former continued accumulation of grid points cumulative distance as the distance from the current frame to match the d (i, j) added, and a current cumulative distance grid. This calculation has reached the grid (n, m), and D (n, m) output, as a result of template matching.)

增加了遗忘因子的最小二乘递推算法，算法更加准确(The least squares recursive algorithm of forgetting factor is added, and the algorithm is more accurate)

Multivariate Gaussian Mixture Model Optimization by Cross Entropy

SampleGenerate.m 生成EM_data数据 GMM_EM_Main.m 执行EM temp2是测试如何求高斯密度的函数(SampleGenerate.m GMM_EM_Main.m temp2)

维纳滤波用于噪声抑制，有辅观测数据和无有辅观测数据的维纳噪声抑制器(Wiener filter for noise suppression, auxiliary data and without auxiliary Wiener noise suppressor observation data)

美国F-16飞机的配平程序、小扰动线性化程序、全量方程求解时域响应、小扰动线性化方程求解时域响应。用MATLAB2014a编写。(The United States F-16 plane balancing procedure, the small perturbation linearization procedure, the full amount of solving the equation of time domain response, small perturbation linearization equation is solved in time domain response. Written by MATLAB2014a.)