PROJECT TITLE :

Improved Bag of Feature for Automatic Polyp Detection in Wireless Capsule Endoscopy Images

ABSTRACT:

Wireless capsule endoscopy (WCE) needs computerized method to reduce the review time for its massive image information. In this paper, we tend to propose an improved bag of feature (BoF) technique to help classification of polyps in WCE pictures. Rather than utilizing one scale-invariant feature remodel (SIFT) feature within the traditional BoF method, we extract totally different textural options from the neighborhoods of the key points and integrate them along as artificial descriptors to hold out classification tasks. Specifically, we tend to study influence of the quantity of visual words, the patch size and different classification strategies in terms of classification performance. Comprehensive experimental results reveal that the simplest classification performance is obtained with the integrated feature strategy using the SIFT and the complete local binary pattern (CLBP) feature, the visual words with a length of one hundred twenty, the patch size of eight*eight, and the support vector machine (SVM). The achieved classification accuracy reaches 93.a pair ofpercent, confirming that the proposed scheme is promising for classification of polyps in WCE images.