The std function has a bias correction which the kurtosis doesn't, if you remove it you get the same result:
rng(31415)
A = randn(7,5);
moment(A,4)./(std(A,1)).^4 % the ,1 in the standard deviation removes the bias correction
kurtosis(A)
ans =
1.4861 1.9326 1.2991 2.0221 3.1096
ans =
1.4861 1.9326 1.2991 2.0221 3.1096
Calculating the other central moments should be then trivial, you just have to known how you want define your standard deviation.
