A method for automatic virtual endoscopy navigation, including: (a) using a fisheye camera to generate an endoscopic image and a depth image from a current position of the camera in lumen computed tomographic (CT) data (b) segmenting a first region and a second region from the depth image, wherein the first region identifies a view direction of the camera and the second region is an area through which the camera can be moved without touching an inner surface of the lumen (c) moving the camera from the current position, while pointing the camera in the view direction, to a next position in the second region and (d) repeating steps (a-c) in sequence using the next position in step (c) as the current position in step (a).
