Informative frame classification method for real-time analysis of colonoscopy video

The proposed classification method in this paper was introduced to classify the images extracted from colonoscopy videos into two categories: informative frames and non-informative frames. The classification method consists of two steps. In the first step, the isolated pixel ratio (IPR) can distinguish the images into three categories: informative frames , non-informative frames and ambiguous frames. In the second step, the total pixels were used to identify the ambiguous frames into either informative or non-informative frames. After the input image frames have been classified into two categories, the non-informative ones will be discarded before being sent to the surgeon for diagnosis. This will reduce the number of frames to be transmitted over distance between the surgeon side and the patient side in tele-surgery which will enhance quality of transmission and achieve higher speed for video transmission.