Feature extraction surf, storing points

I have some number of images that contains faces. First, I am collecting features from all images and need to store features for further purposes. What is the best way to store features? Later, I am gonna use them to match input images to detect face.

face = rgb2gray(imread('...'));
facePoints = detectSURFFeatures(face);
features = selectStrongest(facePoints, 100);

Answer 1

[Turlough Hughes]

One way is to store as a struct, for example

s(ii).features=features; %where ii would be incrementing as you iterate through each image (i.e. for loop)

You can access the features of a given image afterwards as:

features=s(ii).features;

This way of storing your data will be particularly useful if you have other information you wish you wish to store that relates to each image such as:

s(ii).ImageName=ImName;
s(ii).folder=folder;
s(ii).otherdata=otherdata;

If you are not very familiar with structures I recomend reading the following documentation and working through some of the examples.

[Ömer Faruk Kurular]

I want to store them persistently.

[Image Analyst]

Do you mean like to disk in a .mat file by using the save() function?

[Ömer Faruk Kurular]

anythin but persistently? I could store in txt file but would not be effective so looking for a neat way.

[Ömer Faruk Kurular]

It would be so good if I could store structure array persistently.

[Turlough Hughes]

Ah ok, you can save the struct as follows:

for ii=1:n_imgs
    %load image
    features= %etc
    s(ii).features=features;
end

save('s.mat','s')

this will save the struct in your working directory as a file called s.mat.