A new structure tensor based image inpainting algorithm

A new structure tensor based image inpainting algorithm STIA is proposed for solving the deficiencies of the classical Criminisi method, such as the error repair accumulation, high time complexity caused by the unreasonable design of the patch priority, inaccuracy criterion and its global search strategy. Firstly, considering the characteristic of the structure tensor in describing image structure, we make a combination between the structure tensor and the priority function to optimise the repair order of the damaged patches. And this new design idea increases the influence of structural information on repairing. Secondly, to lower the time complexity, an improved matching constraint equation and optimal matching criterion are presented using characteristic value of the structure tensor. Finally, damaged gray and colour images are tested respectively to verify the new algorithm. Experiments show that the improved algorithm not only maintains the structure and texture of images but also makes progress in both subjective visual effect and objective indexes compared with some of the typical image restoration algorithms proposed in recent years.

[1]  Tony F. Chan,et al.  Nontexture Inpainting by Curvature-Driven Diffusions , 2001, J. Vis. Commun. Image Represent..

[2]  Qiuqi Ruan,et al.  Object Removal By Cross Isophotes Exemplar-based Inpainting , 2006, 18th International Conference on Pattern Recognition (ICPR'06).

[3]  Irfan Essa,et al.  Texture optimization for example-based synthesis , 2005, SIGGRAPH 2005.

[4]  Patrick Pérez,et al.  Region filling and object removal by exemplar-based image inpainting , 2004, IEEE Transactions on Image Processing.

[5]  Tony F. Chan,et al.  Mathematical Models for Local Nontexture Inpaintings , 2002, SIAM J. Appl. Math..

[6]  Hailing Zhou,et al.  Adaptive patch size determination for patch-based image completion , 2010, 2010 IEEE International Conference on Image Processing.

[7]  Zhu Kun.  An improved image inpainting algorithm based on texture synthesis , 2011 .

[8]  Ullrich Köthe,et al.  Edge and Junction Detection with an Improved Structure Tensor , 2003, DAGM-Symposium.

[9]  Christian Jutten,et al.  A new approach in decomposition over multiple-overcomplete dictionaries with application to image inpainting , 2009, 2009 IEEE International Workshop on Machine Learning for Signal Processing.