Open Collections

Abstract

The use of both computer generated images and real video images can be made much more effective by merging them, ideally in real time. This motivation is at the basis of Computer Augmented Reality (CAR), which involves both elements of computer graphics and computer vision. This thesis is concerned with an important aspect of CAR: to obtain geometric informa tion about the light sources and about the surface normals from the image pixel values, and use that information to shade the computer generated objects and reshade the real objects when necessary. To acquire light source direction and surface orientation from a single image in the absence of prior knowledge about the geometry of the scene, we assume that the changes in surface orientation are isotropically distributed. This is exactly true for all convex objects bounded entirely by gradually occluding contours, and approximately true over all scenes. This thesis develops an improved method of estimating light source direction and local surface orientation from shading information extracted from pixel values under such assumptions. First and second derivatives of intensity at each pixel are used to compute these estimates, and a weighted sum of all estimated illuminant directions is used for the whole image. We tested our algorithm with different kinds of images: synthetic images and real video images, images with various non-planar shapes, and images with different (non-diffuse) surface reflections. We found that our illuminant direction estimator is able to produce useful results in all cases, and that our surface orientation estimator is able to give useful information in many cases. The main use for such information in the context of CAR is to reshade objects in the real images according to new lighting information, and our tests show that our method is effective in such cases, even when both the light direction and the surface normals have been estimated from the image.