Integral imaging is a technique capable of displaying 3–D images with continuous parallax in full natural color. It is one of the most promising methods for producing smooth 3–D images. Extracting depth information from integral image has various applications ranging from remote inspection, robotic vision, medical imaging, virtual reality, to content-based image coding and manipulation for integral imaging based 3–D TV. This paper presents a method of generating a depth map from unidirectional integral images through viewpoint image extraction and using a hybrid disparity analysis algorithm combining multi-baseline, neighborhood constraint and relaxation strategies. It is shown that a depth map having few areas of uncertainty can be obtained from both computer and photographically generated integral images using this approach. The acceptable depth maps can be achieved from photographic captured integral images containing complicated object scene.
© 2007 IEEEPDF Article