I want to know the what constitutes of a feature vector when a gabor filter is applied to an image for feature extraction? I want feature vector for classification of facial expressions. I have codes for gabor filters and I am getting the filtered images also but i am confused about the feature vector.
1 Answers
1
votes
After you implement the convolution operation for original image and Gabor filter, the complex convolution result in frequency domain can be decomposed to magnitude and phase image based on the real and imag part. The phase responses vary significantly even for spatial locations only a few pixels apart, Gabor phase features are considered unstable and are usually discarded. The magnitude responses, on the other hand, vary slowly with the spatial position, and are thus the preferred choice when deriving Gabor filter based features.
Then you reshape your magnitude image to a long vector, use all the constructed vectors from training database to form a large matrix, apply principal component analysis or linear discriminant analysis to reduce the data dimension, and do the further process implementations.
