A training based Support Vector Machine technique for blood detection in wireless capsule endoscopy images

Wireless capsule endoscopy (WCE) is a non-invasive technique which could detect variety of abnormalities in the small bowel. However, in many cases it is difficult for doctors to distinguish obscure gastrointestinal bleeding of patients; moreover, the diagnosis process of the obtained video could take long time due to the huge number of generated frames. This project provides a method to automatically detect bleeding areas in the WCE images by using Support Vector Machine (SVM) classifier as the main engine with learning based mechanism in order to increase the accuracy. The experiment results show that this could not only advance the accuracy of WCE diagnosis, but also reduce the diagnosis time effectively. To further improve the performance of the proposed approach the length of the learning-based database was also tuned.