User Guided Image Abstraction for Vectorization
D. Çoban (TU Delft - Electrical Engineering, Mathematics and Computer Science)
A.D. Parakkat – Mentor (TU Delft - Computer Graphics and Visualisation)
Elmar Eisemann – Mentor (TU Delft - Computer Graphics and Visualisation)
A Hanjalic – Graduation committee member (TU Delft - Intelligent Systems)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Image abstraction plays an important role in converting pixelated or low quality images into vector images. This paper presents a systematic user interactive process to abstract a generic photograph and to produce suitable content for vectorization. The goal of abstraction is to simplify the content of the photograph while retaining important information, creating a stylized and artistic image as a result. The system offers the user to create an importance map, by selecting regions with high importance or salience. Then, two different methods, bilateral filtering with k-means clustering and mean-shift segmentation, can be used as abstraction methods. Various parameters for different parts of the image and for each method, are exposed to the user to extend the spectrum of variety for the end result, and to tune the final image as desired. Canny's edge detection process is used to create a line drawing from the original image. The user can adjust thresholds, delete edges, and change the thickness and smoothness of edges to alter the line drawing. The resulting images are greatly diverse in terms of the number of colors and the amount of edges they can potentially have. In the end, this method proposes a new interactive way to create images suitable for vectorization, by combining and altering previous abstraction techniques.