A Capsule Robot Attitude Transformation Perception Method based on Intestinal Fold Features

As the world medical level of ascension, the traditional endoscopy has can't satisfy people's needs, using tiny robots enter human body bowel check is becoming a new way of examination. Active robot has become a research hotspot due to its controllable attitude. In this paper, an attitude localization method based on the features of intestinal folds is proposed, and the contour of intestinal tract is obtained by light compensation and adaptive threshold segmentation. The contour HU moment is used for coarse matching, and the shape context is used for fine matching to obtain the matching relationship of contour point pairs. The constraint equation is constructed from the contour centroid and the minimum enclose rectangular centroid angle to limit the search range of point pairs. Finally, the attitude transformation matrix of the robot is solved by using the optimal matching relation and multi-view geometric constraints.