Hello. I can extract SURF features from images using matalb but I can't perform classification on group of images using SURF.
Could anyone tell me how I can use matlab to classify images depend on extracted SURF features?
Thank you!
I think you can use SVM for classification.
These links may be helpful:
http://www.vlfeat.org/overview/tut.html
http://www.robots.ox.ac.uk/~vgg/share/practical-image-classification.htm
What type of images are you dealing with and what kind of classification you want? Please tell us something more about so that we can make meaningful suggestions.
Dear Dr Parul,
i am still in beginning so any database will be welcome.
I try to apply LDA OR SVM classification
i worked on birds database in the following link
http://www-cvr.ai.uiuc.edu/ponce_grp/data/
Thanks
Hi
You may find many classifier for classification purpose in image processing.
After feature extraction next task is to representation of those images i.e creating knowledge base of those images. Finally you divide your data set into testing data set and training data set .
For training data what u hv already done is enough( feature extraction) , then match the images from u r testing class images to training class.
simple classifier is Knn.
the below link is useful ...
http://www.jars1974.net/pdf/12_Chapter11.pdf
Dear Ahmed,
What do you want identify in the image?, How many images do you have?
Do you have a control image? and Finally, some images have similar behavior?
Please, tell me more about your work
Best regards
Hi Chethankumar Bm,
thanks for chapter but my problem is i dont know how can i construct train and test vector using SIFT featuer
Dear Juan Carlos Martinez E.,
I Have images called birds i get it from the following link
http://www-cvr.ai.uiuc.edu/ponce_grp/data/
i want classify them using SURF , using any classification method (in matlab) my problem is i cant figure how can i construct test and train vector from Extracted Surf features.
thanks
Hello,
Create a matrix for training data with number of samples(birds) as rows and its SURF features as columns. then you use this training data matrix into the classifier function.
For better understanding with explanation see,
http://www.mathworks.com/help/stats/support-vector-machines-svm.html#bsr5o09
You can learn from the Matlab explanation about how to train your dataset for classification.
Rekha Jayaprakash,
I did that but i get error that training matrix must be numeric value
Note I build train matrix every ow represent image and ever coulmn containt descriptor vector.
thank,
May I ask what sort of result/output you got after the SURF extraction?
- Badges
- Science topic