Search code examples
matlabpcamatlab-cvst

Local binary pattern and principal component analysis - Matlab

I have a question regarding local binary pattern and principal component analysis. I understand both the methods separately but I do not know how to combine them.

The Matlab output for LBP features can be a vector back to 1-by- N feature vector as shown in the link: http://www.mathworks.com/help/vision/ref/extractlbpfeatures.html

If I generate LBP features vectors of images (50, for example), how can I perform the PCA? Is this a valid approach to use?

Thanks

Solution

  • Assuming you have M images (say, 50), and N features for each image, you could see that as M points in N dimensions. You could use PCA to reduce your dataset to N-r dimensions, where r is the number of dimensions to be removed.