MEASUREMENT OF DEPTH IMAGE CONSIDERING TIME DELAY

An augmented reality presentation system that generates and presents a virtual image free from any latency from a real space. This system has a position/posture sensor for time-sequentially inputting viewpoint position/posture information, stereo cameras for inputting a continuous time sequence of a plurality of images, and an image processing apparatus. The image processing apparatus detects a continuous time sequence of depth images ID from the continuous time sequence of input stereo images, estimates the viewpoint position/posture of the observer at a future time at which a three-dimensional image will be presented to the observer, on the basis of changes in previous viewpoint position/posture input from the position/posture sensor, continuously warps the continuously obtained depth images to those at the estimated future viewpoint position/posture, and presents three-dimensional grayscale (or color) images generated according to the warped depth images to the observer.