An Improved Weighted-Least-Squares-Based Method for Extracting Structure from Texture
Extracting meaningful structures from textured images is an import operation for further image processings such as tone mapping, detail enhancement and pattern recognition. Researchers have pay attention to this topic for decades and developed different techniques. However, though some existing methods can generate satisfying results, they are not fast enough for realtimely handling moderate images (with resolution \(1920\times 1080\times 3\)). In this paper, we propose a novel variational model based on weighted least square and a very fast solver which can be highly parallelized on GPUs. Experiments have shown our method is possible to operate images with resolution \(1920\times 1080\times 3\) realtimely.
KeywordsTexture Structure Weighted least squares GPU
This paper is supported by the Post-Doctoral Research Center of China Digital Video (Beijing) Limited.
- 1.Tomasi, C., Manduchi, R.: Bilateral filtering for gray and color images. In: Proceedings of the Sixth International Conference on Computer Vision, ICCV 1998, Washington, DC, USA, p. 839. IEEE Computer Society (1998)Google Scholar
- 7.Zhang, F., Dai, L., Xiang, S., Zhang, X.: Segment graph based image filtering: fast structure-preserving smoothing. In: Proceedings of the 2015 IEEE International Conference on Computer Vision (ICCV), ICCV 2015, Washington, DC, USA, pp. 361–369. IEEE Computer Society (2015)Google Scholar
- 11.Xu, L., Yan, Q., Xia, Y., Jia, J.: Structure extraction from texture via relative total variation. ACM Trans. Graph. 31(6), 139:1–139:10 (2012)Google Scholar
- 12.Tan, X., Sun, C., Pham, T.D.: Multipoint filtering with local polynomial approximation and range guidance. In: IEEE Conference on Computer Vision and Pattern Recognition, pp. 2941–2948 (2014)Google Scholar
- 13.Ham, B., Cho, M., Ponce, J.: Robust image filtering using joint static and dynamic guidance. In: Computer Vision and Pattern Recognition, pp. 4823–4831 (2015)Google Scholar
- 15.Nvidia cuda home page. http://www.nvidia.com/object/cuda_home_new.html