matlab - Dimensionality reduction in HOG feature vector -

i found out hog feature vector of following image in matlab.

input image

i used following code.

i = imread('input.jpg'); = rgb2gray(i); [features, visualization] = extracthogfeatures(i,'cellsize',[16 16]);   

features comes out 1x1944 vector , need reduce dimensionality of vector (say 1x100), method should employ same?

i thought of principal component analysis , ran following in matlab.

prinvec = pca(features);   

prinvec comes out empty matrix (1944x0). doing wrong? if not pca, other methods can use reduce dimension?

you can't pca on this, since have more features single observation. more observations, 10,000 presumably, , can pca.

see pca in matlab selecting top n components more detailed , mathematical explanation why case.