[gmm+kde] target tracking simulation based on GMM and KDE kernel estimation of MATLAB

1. Software version

matlab2013b

2. Theoretical knowledge of this algorithm

GMM

       GMM It is a method of using a certain number of small Gaussian functions to approximate the probability density function of a variable , It is a commonly used parametric model in probability estimation , This year it is widely used in the field of pattern recognition .

    A limited GMM The description of is very simple , That is, a mixed model is formed by any convex combination of high-tech distribution functions . about n Random variables in dimensional real space x, utilize GMM The probability density function can be expressed approximately ：

 

        Gaussian distribution , Also known as normal distribution . The great mathematician Gauss （Gauss） And 1809 In, he was studying error theory

Derived from . For random variables x, The probability density is ：