Highlight Separation Using Multiple Images with Intensities and Ranges

In this paper, a method for extracting influences of specular reflection and shade from multiple color images is described. As regards extraction, the di-chromatic reflection model is applied. First, the same point on an object is observed from multiple viewpoints, and the pixel values in a window in which the fixation point is being imaged are projected into RGB color space. The shape of each cluster is linear, and, when one of the images observed the specular reflection, the clusters intersect . By using this phenomenon a specular reflection is extracted. To register each color image, range images are measured. Each range image is measured from the same viewpoint with a color image. In this paper, using image extracted specular and shade as texture data, a color object is reconstructed by the use of computer graphics images.