Supplementary material of the paper entitled "Image vectorization and editing via linear gradient layer decomposition"

Introduction

This page contains three sections: 1) results and evaluation, 2) comparisons, and 3) perceptual study. In section 1) and section 2), we show more results to illustrate the effectiveness of the proposed approach. In section 3), we perform a perceptual study to further assess the effectiveness of our method.

1. Results

This section contains three pages. Please click the links below to enter each page.

Navigation of Results and Evaluation Section
Link Introduction
1 All results This page shows all of our 47 examples, all of which are with man-made region segmentation and are vectorized automatically.
2 Results of specifying opaque layers To improve the flexibility of our method, we allow users to specify some regions as opaque layers. Two examples are shown in this page.
3 Time statistics This page provides the complete time statistics. We provide the processing time (in seconds) of each step of our algorithm.
2. Comparisons

In this section, we provide 32 results compared with Photo2ClipArt [Favreau et al. 2017]. We provide two pages to show the differences between Photo2ClipArt and our method. Please click the links below to enter each page.

Navigation of Comparisons Section
Link Introduction
1 Reconstructed results with differences highlighted This page shows the significant differences of the vectorization results generated by Photo2ClipArt and our method.
2 All layers This page shows an interactive GUI for users to better compare the quality of layers generated by Photo2ClipArt and our method.
3. Perceptual study

In this section, we conducted a perceptual study with 32 participants to further assess the effectiveness of our method.

Please see perceptual study.html page for more details.