Email this article 
Automatic Image Style Transfer Using an Augmented Style Set
- Authors: Ponamaryov V.V.1, Kitov V.V.1,2
-
Affiliations:
- Lomonosov Moscow State University
- Plekhanov Russian University of Economics
- Issue: No 3 (2024)
- Pages: 14-20
- Section: COMPUTER GRAFICS AND VISUALIZATION
- URL: https://ogarev-online.ru/0132-3474/article/view/271637
- DOI: https://doi.org/10.31857/S0132347424030029
- EDN: https://elibrary.ru/QAXHSF
- ID: 271637
Cite item
Open Access
Access granted
Abstract
Image style transfer is an applied task for automatic rendering of the original image (content) in the style of another image (specifying the target style). Traditional image stylization methods provide only a single stylization result. If the user is not satisfied with it due to stylization artifacts, he has to choose a different style. The work proposes a modified stylization algorithm, giving a variety of stylization results, and achieves improved stylization quality by using additional style information from similar styles.
Full Text
About the authors
V. V. Ponamaryov
Lomonosov Moscow State University
Author for correspondence.
Email: valera.pon.vp@gmail.com
Russian Federation, Moscow
V. V. Kitov
Lomonosov Moscow State University; Plekhanov Russian University of Economics
Email: v.v.kitov@yandex.ru
Russian Federation, Moscow; Moscow
References
- Gooch B., Gooch A. Non-photorealistic rendering. CRC Press, 2001.
- Strothotte T., Schlechtweg S. Non-photorealistic computer graphics: modeling, rendering, and animation. Morgan Kaufmann, 2002.
- Rosin P., Collomosse J. Image and video-based artistic stylisation. Springer Science & Business Media, 2012. V. 42.
- Gatys L.A., Ecker A.S., Bethge M. A neural algorithm of artistic style // arXiv preprint arXiv:1508.06576, 2015.
- Huang X., Belongie S. Arbitrary style transfer in real-time with adaptive instance normalization // Proceedings of the IEEE international conference on computer vision. 2017. P. 1501–1510.
- Li Y. et al. Universal style transfer via feature transforms // Advances in neural information processing systems. 2017. V. 30.
- Simonyan K., Zisserman A. Very deep convolutional networks for large-scale image recognition // arXiv preprint arXiv:1409.1556, 2014.
- Deng J. et al. Imagenet: A large-scale hierarchical image database // 2009 IEEE conference on computer vision and pattern recognition. IEEE. 2009. P. 248–255.
- Phillips F., Mackintosh B. Wiki Art Gallery, Inc.: A case for critical thinking // Issues in Accounting Education. 2011. V. 26. № 3. P. 593–608.
- Florea C. et al. Pandora: Description of a painting database for art movement recognition with baselines and perspectives // 2016 24th European Signal Processing Conference (EUSIPCO). IEEE. 2016. P. 918–922.
- Buzuloiu V. et al. Adaptive-neighborhood histogram equalization of color images // Journal of Electronic Imaging. 2001. V. 10. № 2. P. 445–459.
- de Villiers J.P. A comparison of image sharpness metrics and real-time sharpening methods with GPU implementations // Proceedings of the 7th International Conference on Computer Graphics, Virtual Reality, Visualisation and Interaction in Africa. 2010. P. 53–62.
- Hassen R., Wang Z., Salama M.M.A. Image sharpness assessment based on local phase coherence // IEEE Transactions on Image Processing. 2013. V. 22. № 7. P. 2798–2810.
- Sanakoyeu A. et al. A style-aware content loss for real-time hd style transfer // Proceedings of the European conference on computer vision (ECCV). 2018. P. 698–714.
- Huang X., Belongie S. Arbitrary style transfer in real-time with adaptive instance normalization // Proceedings of the IEEE international conference on computer vision. 2017. P. 1501–1510.
- Li Y. et al. Universal style transfer via feature transforms // Advances in neural information processing systems. 2017. V. 30.
Supplementary files
Supplementary Files
Action
1.
JATS XML
2.
Fig. 1. Application of style transfer using the method [4]. The first style overlapped worse than the second one because of strong incongruity with the content image (in terms of clarity)
Download (392KB)
3.
Fig. 2. VGG network layers used for stylisation
Download (115KB)
4.
Fig. 3. Example of work of the algorithm of styling by a group of found styles with recolouring of the initial style: a) content; b) standard styling of content a) by style e); c) proposed styling by a group of found styles e) and h); d) proposed styling by a group of styles with recolouring g) and i); e) initial style; f) 1st found style; g) style e) recoloured in e); h) 2nd found style; i) style h) recoloured in e)
Download (391KB)
5.
Fig. 4. Example of work of the algorithm of styling by a group of found styles with transfer of the colour scheme of the original style: a) content; (b) standard styling of content a) by style e); c) suggested styling by a group of found styles e), g); d) suggested styling by a group of styles with recolouring e), g); e) custom style; f), g) similar to (e) styles
Download (376KB)
6.
Fig. 5. Operation of the proposed styling algorithm on the group of found styles at different group size k: a) styling by e), k = 1; b) styling by e) and f), k = 2; c) styling by e) - g), k = 3; d) styling by e) - h), k = 4
Download (566KB)
7.
Fig. 6. Operation of the proposed styling algorithm on the group of found styles at different group size k: a) styling by e), k = 1; b) styling by e) and f), k = 2; c) styling by e) - g), k = 3; d) styling by e) - h), k = 4
Download (576KB)
