Bleeding detection in wireless capsule endoscopy images based on binary feature vector

Wireless Capsule Endoscopy (WCE) is a non-invasive way which is getting its popularity in many hospitals for gastrointestinal examination. However, it produces too many images that need physicians to check manually. This is a huge burden for physicians. To solve this problem, an automatic method based on Support Vector Machine is proposed in this paper. A binary feature vector is used to overcome the drawbacks of conventional color histogram, which compares similarity between histograms rather than checks out the existence of a specified pattern. Considering the property of WCE images, a clipped illumination invariant color space is introduced. Experiments demonstrate that the binary feature vector is more effective than histograms in detecting bleeding patterns of WCE images.