边缘信息诱导的图像矢量化算法

An Image Vectorization Algorithm Using Edge Information

光栅图像的矢量化是数字图像处理领域的重要研究内容.在提取光栅图像边缘信息的基础上,提出一种基于边缘几何和色彩信息的图像矢量化算法.算法首先利用Canny算子提取图像的边缘点,并采用数学形态法细化提取的边缘并连接断裂的边缘;然后根据处理后的边缘点几何和颜色信息,运用B样条方法拟合图像的边缘曲线,得到图像的矢量化表示;最后利用中点算法对边缘曲线的B样条表示进行光栅化处理,并运用邻域平均法对边缘的颜色信息进行迭代,根据图像的矢量化信息重建光栅化图像.实验表明,本算法运行稳定,利用该算法得到的矢量化信息所重建的光栅化图像与原图差异较小,适用于灰度图和彩色图像的矢量化.

Image vectorization is an important and essential aspect of digital image processing. In this paper, we present an image vec- torization algorithm based on extracting geometric and color information of edge from image. In our algorithm,edge information is first detected by Canny operator, and then thinned and connected by Mathematical Morphology method. Based on the extracted geometry and color information of edges, our algorithm vectorizes the image edges by using cubic B-spline curves and the vector image is finally obtained. In order to reconstruct raster image from the vector image for display purpose, the midpoint algorithm is employed in our al- gorithm to rasterize the edge curves, and neighborhood averaging method is also used to diffuse the color sources. Experiments show that our algorithm is robust for various images, and only little difference can be found between original image and the reconstructed im- age of the corresponding raster image by our algorithm. The algorithm can be used to deal with both gray and color images.