python - PCA with sklearn. Unable to figure out feature selection with PCA -

python - PCA with sklearn. Unable to figure out feature selection with PCA -

i have been trying dimensionality reduction using pca. have image of size (100, 100) , using filterbank of 140 gabor filters each filter gives me response 1 time again image of (100, 100). now, wanted feature selection wanted select non-redundant features , read pca might way do.

so proceeded create info matrix has 10000 rows , 140 columns. so, each row contains various responses of gabor filters filterbank. now, understand can decomposition of matrix using pca as

from sklearn.decomposition import pca pca = pca(n_components = 3) pca.fit(q) # q 10000 x 140 matrix

however, confused how can figure out of these 140 feature vectors maintain here. guessing should give me 3 of these 140 vectors (corresponding gabor filters contain info image) have no thought how proceed here.

pca give linear combination of features, not selection of features. give linear combination best reconstruction in l2 sense, aka 1 captures variance.

what goal? if on 1 image, kind of selection give features discriminate best parts of image against other parts of same image.

also: garbor filters sparse basis natural images. not expect interesting happen unless have specific images.

python image-processing scikit-learn pca dimensionality-reduction