Adaptive patch based texture synthesis using contourlet transform

Article type: Research Article

Authors: Vinod Kumar, R.S. | Arivazhagan, S.

Affiliations: Department of Electronics & Communication Engineering, Noorul Islam University, Tamil Nadu, India | Department of Electronics & Communication Engineering, Mepco Schlenk Engineering College, Tamil Nadu, India

Note: [] Corresponding author. R.S. Vinod Kumar, Department of Electronics & Communication Engineering, Noorul Islam University, Tamil Nadu, India. E-mail: rsvinodkumar@niuniv.com

Abstract: An adaptive patch based texture synthesis using contourlet transform is discussed in this paper. The synthesis starts with patches of as bigger size as possible. Whenever the error criterion is not satisfied, the patch size is reduced by one fourth and the synthesis is continued. In the proposed work, the patch satisfying the error criterion using contourlet transform is selected as the candidate patch. Generalized mathematical expressions have been developed to select the patches of the required sizes. The experimentation is carried out with different contourlet filters and based on the quality of the synthesized outputs; a specific level of decomposition is identified. The quality of the obtained outputs is satisfactory.

Keywords: Image processing, texture synthesis, euclidean norm, minimum cost path, contourlet transform

DOI: 10.3233/IFS-141389

Journal: Journal of Intelligent & Fuzzy Systems, vol. 28, no. 3, pp. 1061-1070, 2015