Reduction of redundant frames in active wireless capsule endoscopy

Abstract

Wireless capsule endoscopy (WCE) is a useful technique to detect the abnormal lesion such as bleeding and polyp location in human gastrointestinal (GI) tract. It also helps in examine the ulcer and cancer in gastrointestinal tract. Active WCE is a new technique to cope up with the problem of nonuniform motion of passive WCE. There are several numbers of similar frames in active WCE video due to natural peristalsis in gastrointestinal tract of the human body. Here, a reduction model for active WCE video for the reduction of visualization time by a clinician is proposed. This reduction model consists of two levels. In the first level, we apply intensity based skip-prediction method to find out duplicate frames. In the second level, random sample consensus (RANSAC) combined with Harris corner detection is applied to the remaining redundant frames.