Fast Feature-Based Metamorphosis and Operator Design Tong-Yee LEE 1 , Young-Ching LIN, Leeween LIN 2 , Y.N. SUN Department of Computer Science and Information Engineering National Cheng-Kung University Tainan, Taiwan, R.O.C Abstract Metamorphosis is a powerful visual technique, for producing interesting transition between two images or volume data. Image or volume metamorphosis using simple features provides flexible and easy control of visual effect. The feature-based image warping proposed by Beier and Neely is a brute-force approach. In this paper, first, we propose optimization methods to reduce their warping time without noticeable loss of image quality. Second, we extend our methods to 3D volume data and propose several interesting warping operators allowing global and local metamorphosis of volume data. 1 Corresponding author: tonylee@mail.ncku.edu.tw 2 Institute of Computer and Information Engineering, National Sun Yat-Sen Univ., Kaohsiung, Taiwan, ROC 1. Introduction Metamorphosis or warping technique is a powerful tool to transform one image into another or generate new 3D model from a given one. Using this technique, many exciting visual effects in film and television is realized by animating complex models and changing their physical attributes such as shapes and positions. In general, 2D warping can be achieved by generating 2D images from 3D metamorphosis. Levoy et al [1] discuss the shortcomings of 2D warping and thus suggest 3D morphing. In the past, there have been many efforts done in 2D and 3D metamorphosis. Lee et al present a survey of 2D warping algorithms including those based on mesh morphing, field morphing, radial basis functions, thin plate splines, and energy minimization [2]. Prior work on field morphing [3] will be discussed in section 2. For mesh morphing [4], two meshes from the source and target images are used to define the spatial transformation that maps all points in the source image onto the target image. There is no folding or discontinuity allowed on mesh morphing. The radial basis functions and thin plate splines use more general form of features such as lines and curves [5]. On energy minimization method, the morphing transition is specified by physically meaningful energy terms and satisfied by minimizing their sum [6]. It results in a natural warping but with very high computation cost. Several prior researches on volume morphing were presented in the past. Levoy et al extend [3] to 3D volume data and present optimization for computational efficiency. [7] attempts to automatically generate morphing without the aid of user input. Some work [7,8] transforms the volume data into frequency domain to perform morphing. Recently, Yagel et al [9] use ray deflector technique for volume morphing and allow local control of shape deformation. This approach computes morphing in those regions of volume data that contribute to the final image. Beier and Neely propose a feature-based (field) metamorphosis method [3]. Using this method, an animator begins with establishing correspondence with pairs of feature primitives such as points and line segments between two images or models. In this paper, we attempt to optimize this method. The main contributions of this paper are as follows. First, we present two methods to improve [3] by taking into account the positions of features. The fastest of two proposed schemes achieves 20 times faster for our test images. Second, we extend the fastest method to 3D volume data as did in [1]. However, our extension allows both global and local control of volume deformation. Additionally, we describe several interesting morphing operators similar to