gmm_evaluate.m

来自「用粒子滤波算法进行跟踪的matlab代码。」· M 代码 · 共 13 行

function w = gmm_evaluate(g,x)
%
% assumes weights are not log-likelihood
% TODO: implement the fast gauss transform (FGT)

N = size(x, 2);
w = zeros(1, N);

for i=1:size(g.x, 2)
    dx = x - repcol(g.x(:,i), N);
    w  = w + g.w(i)*gauss_likelihood(dx, g.P(:,:,i));
end