基于区域纹理合成的图像修补算法

付绍春; 楼顺天

doi:10.3724/SP.J.1146.2008.00632

基于区域纹理合成的图像修补算法

doi: 10.3724/SP.J.1146.2008.00632

基金项目:

国家自然科学基金(60775013)资助课题

计量
- 文章访问数: 2950
- HTML全文浏览量: 88
- PDF下载量: 2193
- 被引次数: 0
出版历程
- 收稿日期: 2008-05-19
- 修回日期: 2008-10-20
- 刊出日期: 2009-06-19

Regional Texture Synthesis Based Image Inpainting

摘要

摘要: 针对传统修补技术在图像恢复中出现的纹理模糊现象，该文根据纹理合成技术，提出了一种区域纹理合成的修补算法。算法根据破损点邻域特征选择模板，去除模板内的破损点后进行匹配，在搜索范围内寻找最佳匹配点。该文采用由边缘向中心逐步推进的修补次序，提高破损区域内纹理结构延伸方向的正确性以及连续性，避免出现纹理模糊现象。算法能够适用于多种性质的纹理，在处理真实图像时使人在视觉上获得满意的效果。
- 图像修补;纹理合成;模板选择;匹配
Abstract: For solving the problem of texture blurring brought by traditional inpainting method, a regional texture synthesis based inpainting algorithm is proposed. The feature in the adjacent domain of each broken pixel is analyzed for choosing the template, and the broken pixels should be excluded from the template. The optimal solution for the broken pixel is determined by matching the template in searching area. The inpainting process proceeds from the edge to the center, for improving the validity and continuity of propagating direction of the texture in the flawed region, meanwhile the blurring can be eliminated. The algorithm is available to a wide variety of real-world textures, and produces a satisfying result for human visual system.

HTML全文

参考文献(1)

Bertalmio M, Sapiro G, and Caselles V, et al.. ImageInpainting. Proceedings of the 27th annual conference onComputer graphics and interactive techniques, New Orleans,2000: 417-424.[2]Chan T and Shen J. Mathematic method for local non-textureinpainting[J].SIAM J. Appl. Math.2001, 62(3):1019-1043[3]Chan T and Shen J. Non-texture inpainting by curvaturedrivendiffusions (CDD) [J].Visual Communication andImage Representation.2001, 12(4):436-449[4]Criminisi A, Perez P, and Toyama K. Region filling andobject removal by exemplar-based image inpainting[J].IEEETrans. on Image Processing.2004, 13(9):1200-1212[5]Sun J, Yuan L, and Jia J, et al.. Image completion withstructure propagation[J].ACM Trans. on Graphics.2005,24(3):861-868[6]Ballester C, Bertalmio M, and Caselle V, et al.. Filling-In byJoint Interpolation of Vector Fields and Gray Levels[J].IEEETrans. on Image Processing.2001, 10(8):1200-1211[7]Bertalmio M. Strong-continuation contrast-invariantinpainting with a third-order optimal PDE[J].IEEE Trans.on Image Processing.2006, 15(7):1934-1938[8]Efros A and Freeman W. Image quilting for texture synthesisand transfer. Proceedings of the 28th annual conference onComputer graphics and interactive techniques, Los Angeles,2001: 341-346.[9]Kopf J, Fu C W, and Cohen-Or D, et al.. Solid texturesynthesis from 2D exemplars[J]. ACM Trans. on Graphics,2007, 26(3): 2.[10]Zhou K, Huang X, and Wang X, et al.. Mesh quilting forgeometric texture synthesis[J].ACM Trans. on Graphics.2006,25(3):690-697[11]Wei L Y and Levoy M. Fast texture synthesis usingtree-structured vector quantization. Proceedings of the 27thannual conference on Computer graphics and interactivetechniques, New Orleans, 2000: 479-488.[12]Efros A and Leung T K. Texture synthesis by non-parametricsampling. Proceedings of the 7th IEEE InternationalConference on Computer Vision, Kerkyra, Greece, 1999:1033-1038.[13]Wen L, Zhang D P, and Liu Z Y, et al.. A parallel algorithmfor image information restoration. Proceedings of the 4thInternational Conference on High Performance Computing inthe Asia-Pacific Region, Beijing, 2000: 790-793.[14]Bertalmio M, Vese L, and Sapiro G. Simultaneous structureand texture image inpainting [J]. IEEE Trans. on ImageProcessing, 2003, 12(8): 882-889.[15]Tauber Z, Li Z N, and Drew M. Review and preview:Disocclusion by inpainting for image-based rendering[J].IEEE Trans. on Systems, Man, and Cybernetics, Part C:Applications and Reviews.2007, 37(4):527-540

施引文献

资源附件(0)

访问统计