Image vectorization processing method based on boundary extraction and color reconstruction

Abstract

The invention relates to an image vectorization processing method based on boundary extraction and color reconstruction, and the method comprises the steps: calculating the edge saliency of each pixel in a to-be-processed raster image based on gradient information, and obtaining the edge saliency of each pixel in the to-be-processed raster image according to the adjacent relation between all initial boundary pixels of the to-be-processed raster image; connecting the central points of the initial boundary pixels to form a curve, taking the curve as an initial boundary curve of the vectorized image to be generated, then calculating symmetric energy and rigid energy of edge saliency mapping about the initial boundary curve, and further constructing an initial boundary curve optimization objective function based on the obtained symmetric energy and rigid energy, so as to generate the vectorized image. The method comprises the following steps of: acquiring an optimal boundary curve, generating an adaptive triangular grid of a raster image based on the optimal boundary curve, and finally performing image color coding by taking the generated adaptive triangular grid as a geometric domain to obtain a vectorized image, so that the boundary curve of the vectorized image is straighter and smoother, and the color change of the image is more natural.

Description

technical field [0001] The invention relates to the field of image processing, in particular to an image vectorization processing method based on boundary extraction and color reconstruction. Background technique [0002] In the field of image processing, vector graphics are a common processing object. The colors and shapes (including points, lines, and curves) of vector graphics are given explicitly or implicitly by mathematical formulas. Vector graphics maintain clarity without distortion even when enlarged many times, while raster images become pixelated and jagged, lose clarity, and are very prone to distortion when enlarged. Based on the above characteristics of vector graphics, image vectorization is to convert raster images into vectorized images, so that the converted images have the advantages of vector graphics, and then make image vectorization technology more effective in color editing, image embedding and image reconstruction. applications are becoming more an...

AI Technical Summary

Problems solved by technology

[0004] First of all, the existing image vectorization processing methods are difficult to accurately find the segmented smooth curve of the jagged raster edge (or called the raster boundary) of the raster image, especially when the color gradient is not obvious, to find the segmented smooth curve The difficulty will increase;

[0005] Secondly, most of the existing image vectorization processing methods are implemented based on empirical curve fitting technology, and do not take into account the observer's visual perception of raster images

[0006] In addition, since the edge of the detected raster image is essentially a sudden change in color or light intensity, the raster image will degenerate relative to the original condition. The existing image vectorization processing methods usually focus on the boundary of the raster image Both regions and non-border areas are smoothed with fidelity, which will cause the change of border color to not be obvious across

Images