2D Shape Blending Based On Visual Feature Decomposition And Its Applications

2D shape blending is a visually smooth transformation from source shape to destination shape. It has been widely used in 2D character animation, pattern matching, and geometric modeling. There are two problems in 2D shape blending: vertex correspondence and vertex interpolation path. Since 1980's, many algorithms were proposed to achieve smooth 2D shape blending. In this dissertation, the concept of visual feature decomposition is proposed. It is applied to 2D shape blending without or with texture. As the extension and application of above feature-based 2D shape blending algorithms, a weighted 2D Copy-and-Paste operation is proposed for 2D shapes design and multiple 2D shapes blending. This paper is organized as the following: In Chapter 1, the concept and contents of 2D shape blending are introduced. The state of art about 2D shape blending is given then. In Chapter 2, three feature-based 2D shape blending methods are proposed in detail. First, a visual feature-based 2D shape blending method is presented. Then the direction axis of visual future is defined, and an approximation extraction algorithm is proposed. The direction axis is applied to 2D shape blending to achieve smooth transition effects with the preservation of original visual features. Finally, combined with image morphing method controlled by line pairs, an extended 2D shape blending is proposed, which can process both shapes and interior textures. A weighted 2D Copy-and-Paste operation is presented in Chapter 3, which is based on multiple 2D shapes blending. By using the Copy-and-Paste operations, user can design new polygonal shapes by smoothly combining the parts from different input 2D shapes. The algorithm can also be applied to generate smooth morphing sequences among multiple 2D shapes. The conclusion is drawn and future research is proposed in Chapter 4.