Prior work on feature-based 2D morphing [2] will be discussed in section 3.
Prior work in volume morphing comprises [9], [8], and [5]. These approaches can be summarized in terms of our warping/blending framework.
[5] and [8] have both presented warping techniques. [5] examined the theory of extending selected 2D warping techniques into 3D. A UI was not presented, however, and only morphs of simple objects were shown. [8] presents an algorithm which attempts to automatically identify correspondences between the volumes, without the aid of user input.
[9] and [8] have suggested using a frequency or wavelet representation of the volumes to perform the blending, allowing different interpolation schedules across subbands. In addition, they have observed that isosurfaces of the morphs may move abruptly, or even completely disappear and reappear as the morph progresses, destroying its continuity. This suggests that volume rendering may be superior to isosurface extraction for rendering the morphs.
Our paper is partitioned into the following sections: Section 2 covers volume acquisition methods. Sections 3 and 4 present the warping and blending steps of our morphing algorithm. Section 5 describes an efficient implementation of our warping method and section 6 discusses our results. We conclude with suggestions for future work and applications in section 7.