Retinex

曲线模型的半自适应交互多模型跟踪方法 我存的一篇论文，拿来与大家共享(When using interactive multiple model (IMM) to track targets,to select smallermodel set to better cover the move- ment property and the maneuvering level of various targets and meanwhile to reduce computation complexity are both worthwhile and also necessary research tasks to project implementation.The main work is done through the IMM algorithm of curvilinear model,the construction of angle velocity estimation model and the extension of it to state equations,whichmake it possible for normal acceleration to be self-adaptively adjusted,hence it reduces the model setto associate onlywithMcurvilinearmodels selatingtotangential acceler- ation.Computation burden is reduced and algorithm’s applying range is extended.The simulation further proves the conclusion )

This is a set of MATLAB m-files implementing the mixture fitting algorithm

Shapiro提出的内嵌零树小波编码算法(EZW Embedded Zerotree Wavelet),其出发点在于小波变换后各级子带系数之间在空间上和方向上所呈现出带间相似性,这种空间相似性可以用一种新型的数据结构--零树(Zerotree).EZW算法就是充分利用这种相似性,获得高性能的图像编码算法.(Shapiro' s embedded zero tree wavelet coding algorithm (EZW Embedded Zerotree Wavelet), whose starting point is at all levels of sub-band wavelet transform coefficients and the direction in space with the similarity between shows, this space can be similar with a new data structure- zero tree (Zerotree). EZW algorithm is to take full advantage of this similarity, access to high-performance image coding algorithms.)

比较了三种学习方法 1 gradient descent 2 steepest descent with line search 3 conjugate gradient method. 三种方法都在一个.m文件中被实现。(compare three learning methods: 1 gradient descent 2 steepest descent with line search 3 conjugate gradient method.)

有限单元法求解一阶偏微分方程，使用的方法是加权余量法，比二阶方程更加可靠(Finite element method for second order partial differential equations, a method using a weighted residual method is more reliable than the first-order equation)

关于一级倒立摆LQR+PID控制。包含源程序，simulink仿真模型等(On the inverted pendulum LQR+PID control. Including source program, Simulink simulation model, etc.)

几篇关于神经网络泛化的论文.几篇关于神经网络泛化的论文(several neural network generalization of the papers. Several of the generalization of the neural network papers)

for embedding text in image

embedding and extraction by DCT

经验模态分解和希尔伯特黄变换源程序，可以直接在matlab里运行(Empirical mode decomposition and Hilbert huang transform source code, can be directly run in matlab)