I need to know what is the best algorithm to perform feature extraction between images have significant changes?
you can use the MFCC (Mel Frequency Cepestral Coefficients) as the best feature extraction algorithm and either the artificial neural network or the Support Vector machine as classifiers.
best wishes
Dear El-Sayed, I thouht that MFCC technic is only suitable for 1D data (like sound stream). Is where any work that describing MFCC aplication for image processing?
Dear Valery
Any two dimensional image can be transferred to one dimensional data and hence you can apply the MFCC as a feature extraction algorithm.
You can check PLBP (pyramid of local binary pattern) features, which are proved to extract discriminant features from images.
I have used PCA for features in unsupervised change detection and it is simple.
Thanks a lot dears Rizawan and Stuti. but that will not overcome the image have significant changes
Sir, I have recently used FAST, SURF and SIFT features with Optical Flow HS, LKT and Correlation OF to identify changes in the consecutive video frames. I observed good performance using these methods. And I like the idea of Dr.El-Sayed Mahmoud El-Rabaie for image feature extraction. I will definitely try it. Thank You Sir.
Dear Polurie, am very thankful for your reply. however, transfer the 2D image to frequency domain will not be good idea. hence our imagery have significant differences in rotation, scaling, starching and skewing so these algorithms will fail or produce poor result as we tried after applied SIFT and SURF.
- Badges
- Science topic
- Similar topics
- Computer Science
- Images