Loading...

Applications of Sparse Representation in Digital Image Inpainting

Dehghani Tafti, Zahra | 2019

477 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 52391 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Babaiezadeh, Massoud
  7. Abstract:
  8. Image inpainting is the process of reconstructing lost parts of damaged images based on collected local information, or even general information, as prior knowledge. The image inpainting’s objective is to improve the damaged images, for example restoring missing pixels caused by folding, erasing background’s text in an image, removing watermarks from an image, or editing an image such as eliminating an object or a person from the image. The majority of the image inpainting algorithms approaches the problem by signal restoration from remaining samples or iterative methods to complete the damaged images. Algorithms based on samples, algorithms based on partial differential equations, and algorithms based on spares representation are some instances from the image inpainting algorithms. Sparse representation of signals has been used in signal processing fields such as image inpainting, image denoising, and pattern recognition. In the sparse representation of signals, the least sufficient signals are chose from many basis signals to represent a signal. Each basis signal is called an “atom” and the set of basis signals is called a “dictionary”. In the image inpainting algorithms based on sparse representation each image patch is considered as a signal, and sparsity of these signals in an unknown dictionary is assumed, then incomplete patches are restored by applying dictionary learning methods. Dictionary learning algorithms iterate between two essential steps, the first is “sparse representation”, and the second is “dictionary update”. In this thesis, we firstly review sparse representation and dictionary learning algorithms, then we introduce the image inpainting problem and its algorithms. Afterward we discuss about the image inpainting algorithms based on sparse representation more precisely. It should be noted that sparse representation coefficients and dictionary have to improve in every step of these algorithms, i.e., error has to decrease and the result converges to intact image. To address this problem, we apply dictionary learning problem convexing methods on the image inpainting algorithms, and we discuss proposed method results based on curves, tables, and restored images for different masks, and different images. Finally, we propose an image inpainting algorithm by using intact patches in the damaged image, and dictionary learning algorithms
  9. Keywords:
  10. Sparse Representation ; Dictionary Learning ; Image Inpainting ; Image Processing

 Digital Object List

 Bookmark

No TOC