number-recognition

说明：  这是个关于图像分水岭分割算法的源程序，用matlab实现。(This is a watershed segmentation algorithm on the image source, using matlab to achieve.)

ukf仿真滤波性能分析,针对目标跟踪算法的滤波(UKF algorithm anlysis for target tracking)

说明：  利用平均声强器法测定目标方位，经验证，估计方位与假设方位一致(The average sound intensity using target orientation determination device, proven, estimates and assumptions bearing the same direction)

An unsuperwised SOM classificatoin on the wheather data of a metropolis over 40 years

高阶谱分析，很实用，能帮助分析非平稳信号(Higher order spectral analysis, it is useful to help analyze non-stationary signals)

广义互相关时延估计，瑞利限广义互相关时延估计，瑞利限广义互相关时延估计，瑞利限(Generalized cross-correlation delay estimation, generalized Rayleigh limit cross-correlation delay estimation, generalized Rayleigh limit cross-correlation delay estimation, the Rayleigh limit)

说明：  最近看到有人提问怎样用FFT实现频谱分析，我认为能够进行到频谱分析的人，对于系统研究的深度基本上已经相当不错了。本人今天做了一个例程，发给大家参考一下。 历程主要内容(signal_analyzer.rar)如下： test1.txt 第一列为时间t，0.001s为步长；第二列为系统的输入信号；第三列为系统的输出信号。 signal_analyzer.m首先将test1.txt 的数据读入workspace，然后进行FFT，分析出两信号的频率，输出对于输入的幅值衰减（db），输出对输入的相位差。(The question was asked recently to see the realization of how to use FFT spectral analysis, I think that spectral analysis can be carried out to those who study the depth of the system basically is already quite good. Today I made a routine, distributed to Members for reference. The main contents of the course (signal_analyzer.rar) are as follows: test1.txt the first time as t, 0.001s for the step the second as the system input signal the third as a system output signal. signal_analyzer.m first test1.txt data read into workspace, and then proceed to FFT, analysis of the two signal frequency, output amplitude for input attenuation (db), the output of the input phase difference.)

基于高斯线性的概率密度函数滤波估计，以此为蓝本有很多改进或类似的算法，比如CDKF-PHD等(Based on Gaussian probability density function of the linear filter estimation, there are a lot of improvements as a blueprint or similar algorithms, such CDKF-PHD, etc.)

先做灰階圖片，再利用IMHIST來做統計，在做SMOOTH來讓使用者更容易了解圖片的特色 (Grayscale images do first, and then use IMHIST do statistics, SMOOTH do to make it easier for users to understand the characteristics of the picture)

NUFFT最原始的资料，让你更清晰的了解NUFFT算法。(NUFFT original presentation, so you clearly understand NUFFT algorithms.)