How to do pixel based classification using SVM classifier
8 Ansichten (letzte 30 Tage)
Ältere Kommentare anzeigen
Good day, I am graduate student and currently working on retinal blood vessel segmentation. I am new to this. I am supposed to classify which pixel is vessel pixel and which is not. I want to use SVM as a classifier for pixel based classification task. As I am a beginner, I don't know how may I train my classifier for this task and how may I get the segmented blood vessel network as an output. I am using drive database for this purpose its publicly available. I would like to seek some professional advice from the community to point me in the right direction on what I should be looking at or doing. I will be very grateful for your valuable suggestions. Regards, Arslan
0 Kommentare
Antworten (1)
Image Analyst
am 25 Nov. 2013
The two classes you want are "blood vessel" and "fundus". Okay, fine. Now, what are the two features that you hope to identify clusters (classes) in? https://en.wikipedia.org/wiki/Support_vector_machine Gray level? Alright, seems reasonable. What else? Variation in the window? Possibly. Did you try stdfilt() or entropyfilt()? Or do you have some other feature? Did you check section 20.5 here http://iris.usc.edu/Vision-Notes/bibliography/contentsmedical.html#Medical%20Applications,%20CAT,%20MRI,%20Ultrasound,%20Heart%20Models,%20Brain%20Models for algorithms that segment out vessels from ophthalmological images?
4 Kommentare
Image Analyst
am 25 Nov. 2013
Personally I haven't used SVM, but I would guess, based on Wikipedia, that you just take a representative sample that spans all cases you're ever likely to see and run it through to get a model. Then you use that model on all subsequent data. You might look through the stats tutorials here http://www.sportsci.org/resource/stats/
Siehe auch
Kategorien
Mehr zu Recognition, Object Detection, and Semantic Segmentation finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!