【作者】 李学营；

【导师】 苏洲；

【作者基本信息】 河海大学 ， 机械制造及自动化， 2007， 硕士

【摘要】 随着计算机技术的发展，各种CAD软件得到广泛的应用。目前，在机械设计以及其它设计领域，大多数的系统都是由计算机辅助设计完成。但是，输入到计算机的图像文件需要保存每一个像素点，因此需要占用大量存储空间。为了减少图像信息量，就要将点阵图像转换成矢量图形，以节省内存、便于编辑和输出新图形。 论文主要讨论了基于细化的线划图的矢量化方法。首先对灰度图进行二值化处理，然后利用数学形态学的基本运算——膨胀运算和闭运算对二值图进行消噪及去毛刺处理。采用了与模板匹配的细化算法对图像进行细化，用该算法对线状要素进行细化，避免了端点、孤立点等信息的丢失，有效地提高了细化速度；对于多条细化线构成的复杂交叉点，按惯性跟踪规则来处理，既减少了交点的偏移量，又保证了所识别线段的完整性。在总结已有算法的基础上，把稀疏矩阵与基于细化的链码追踪的矢量化方法相结合，有效地减少了数据的存储量，提高了数据储存的效率。 主要由直线构成的图像，经过直线拟合以后可以得到满意的结果，但对于存在二次曲线的图像，如果完全用小段直线拟合，会导致点阵图像与矢量图像的误差较大，精度达不到要求。为了进一步提高图像矢量化的精度，本文在直线拟合的基础上运用曲线拟合的方法实现了二次曲线的矢量化。更多还原

【Abstract】 With the development of computer technology, many CAD softwares are applied widely. At present, most systems of mechanical design and others are accomplished by computer aided design. However, obtained images information is raster information whose data are large because it is need to conserve the data of every pixel. In order to cut down image information, it is necessary to convert raster to vector.The paper mainly discusses vectorization based on thinning of line drawing. Firstly, gray image changed into binary image. Then, noise and stub burr is removed by using basic operation of mathematical morphology—dilation and closing operation. In the course of thinning to target image, the arithmetic of suited cyclostyle is adopted, which prevents the terminals and the acnodes from losing and reducing effectively time in doing so. In order to insure integrality of lines and decrease offset of crossing points, the rule of inertial tracked is used to deal with complex crossing points. Based on the original thinning arithmetic, combine sparse matrix with vectorization of chain code traced, the storage of data is decreased and the efficiency of storage is improved.Image mainly made up of lines has satisfying result after fitting by line, the image that has conic, completely fitting by small lines leads to some difference between the raster and vector. In order to improve the veracity of vectorization, the vectorization of conic is realized by means of curve fitting in this paper.更多还原