Eric的专栏

转载自：http://users.rowan.edu/~polikar/WAVELETS/WTpart4.htmlROBI POLIKARMULTIRESOLUTION ANALYSIS: THE DISCRETEWAVELET TRANSFORMWhy is the Discrete Wavelet Transform Needed? Although the dis

转载自：http://users.rowan.edu/~polikar/WAVELETS/WTpart3.html

ROBI POLIKARFUNDAMENTALS: THE FOURIER TRANSFORM AND THE SHORT TERM FOURIER TRANSFORM FUNDAMENTALS Let's have a short review of the first part. We basically need Wavelet Transform

写在前面：最近刚开始看小波变换，看了各种中文，总感觉解释的不懂，hou'lia

傅里叶变换的一大用途是从混杂的时域信号中找出其中各频率成分的分布。以一个由50Hz、120Hz两个频率正弦信号和随机噪声叠加得到的信号为例（采样频率1000Hz）：A common use of Fourier transforms is to find the frequency components of a signal buried in a noisy time domain s

转载自：http://betterexplained.com/articles/an-interactive-guide-to-the-fourier-transform/

Figure1，是a=0.4*sin(4*w*(x))的图形，Figure2，是b=1.6*cos(12*w*(x))的图形。这两个图形，在时间轴上，很容易看出来。但是两个的和，也就是a+b，如Figure3所示，里面的一些信息就看不出来了。但是做一个傅里叶变换，转换到频域上，如Figure4所示，就很明显了。Figure4的横坐标是频率，纵坐标是幅值，就可以看出Figure3是有两个信号组成的，

The mean shift algorithm is a nonparametric clustering technique which does not require priorknowledge of the number of clusters, and does not constrain the shape of the clusters.[1]Mean shift是一

Akima相关实现的代码：Akima's original paper:``A new method of interpolation and smooth curve fitting based on local procedures'', Journal of ACM 17, 4 (1970), 589-602http://student.ndhu.edu.tw/~u9111023