抽象图像矢量化

发布时间：2018-10-11 09:22

【摘要】：图像抽象化是数字图像处理中的一种重要技术,在卡通动画、计算机艺术创作、二维游戏、广告创意以及互联网等领域具有广泛的应用。如何在不同分辨率的设备之间传输、处理和显示抽象化图像是一个值得关注的问题。基于几何描述的图像矢量化技术具有表示紧凑、所需的存储空间更小、可以进行高质量缩放等优势,成为解决这个问题的一个可能途径。本文在结合形态简化和流线感抽象化技术的基础上,对抽象化图像进行量化处理,提取出其矢量边界数据,并构建相应的矢量图形绘制算法。论文的主要贡献为： 1.提出了一种抽象化图像的矢量化方法。矢量化图像具有抽象化风格,且与图像分辨率无关,减少了信息冗余。算法无需各种繁琐的人工交互操作,可以自动实现。 2.设计了一种基于边界表示的图像矢量化模型。针对抽象化图像中的不同区域,提出了一种新的封闭边界提取方法,且相邻区域之间没有空隙。结合边界信息、区域内色彩信息和泊松扩散,实现了图像的矢量化描述。
[Abstract]:Image abstraction is an important technology in digital image processing. It has been widely used in cartoon animation, computer art creation, two-dimensional games, advertising creativity and the Internet. How to transmit, process and display abstract images between devices with different resolution is a problem worthy of attention. Image vectorization based on geometric description has the advantages of compact representation, smaller storage space, high quality scaling and so on, which is a possible way to solve this problem. Based on the techniques of morphological simplification and streamline abstraction, the abstract image is quantized, the vector boundary data is extracted, and the corresponding vector drawing algorithm is constructed in this paper. The main contributions of this paper are as follows: 1. A vectorization method for abstract images is proposed. Vectorization image has abstract style and is independent of image resolution, which reduces information redundancy. The algorithm does not need all kinds of complicated manual interactive operation, and can be realized automatically. 2. 2. An image vectorization model based on boundary representation is designed. In this paper, a new closed boundary extraction method is proposed for different regions in abstract images, and there is no gap between adjacent regions. Combined with boundary information, color information and Poisson diffusion, the image vectorization is realized.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP391.41

【共引文献】