Dissertation > Excellent graduate degree dissertation topics show

Study of partial differential equations and the image restoration algorithm based on texture synthesis

Author: TianSuYun
Tutor: WangXiaoMing
School: Shaanxi Normal University
Course: Computer Software and Theory
Keywords: image restoration image denoising image inpainting Partial DifferentialEquation(PDE) texture synthesis priority
CLC: TP391.41
Type: Master's thesis
Year: 2013
Downloads: 68
Quote: 0
Read: Download Dissertation

Abstract


The image restoration, which is the foundation of image processing problem, including image denoising and image inpainting, is one of the most basic research contents in image processing. It means a lot in the domain of theory and practice. Since1990s, image restoration algorithm based on partial differential equations which has been developed. Compared with the traditional algorithms, it has a better theoretical basis and stable performance.We introduce the background, significance, researching status at home and abroad and basic theories of image restoration in the first part. We present some classic models of image restoration. And we present two kinds of algorithms which based on partial differential equations in the domain of image denoising of image restoration in this dissertation. First, we analyse the characteristic of the ROF model and harmonical model and calculcate the form of the numerical discretization. Second, by exploiting the local features of the image, the ability of the ROF model in denoising completely, protecting image edges and producing "ladder effect" for processing image smoothing areas, the ability of the harmonical model in denoising incompletely, blurring image edges and overcoming the "ladder effect" for processing image smoothing areas, and the advantages of the Laplacian operator in enhancing edges, the first image denoising model, RHL was proposed. And we designed and implemented the RHL algorithm. Addition, syncretizing the characteristics of the image inpainting model which can remove salt and pepper noise, we proposed the second model RHI on the basis of the proposed RHL model. And we designed and implemented the RHI algorithm. Through a large number of experimental simulation, the results show that the proposed two algorithms, RHL and RHI, have better performance in terms of visually and quantitatively than other algorithms, which combine the advantages of the ROF model and harmonical model in image denoising effectively. Compared with other PDE based algorithms, our proposed two algorithms can better remove noise, protect smooth regions and edge information much better.In this dissertation, we propose two kinds of algorithms, which based on partial differential equations and texture synthesis in the region of image inpainting of image restoration. The traditional image inpainting algorithms based on partial differential equations applies to repair the image with cracks and stains. Compared with the algorithms based on texture synthesis, they have not a satisfactory result for the area of big-scale defect, defect area width is greater than repaired area width and containing rich texture information. To solve these problems, image inpainting algorithms based on texture synthesis is better able to repair these areas. Therefore, in this paper, we present an improved adaptive image inpainting algorithm based on texture synthesis. This algorithm performs a task through the best-first inpainting strategy based on priority defined to each boundary pixels. The priority computation is biased toward these areas which contain a wealth of know information and structure information. So that these areas with texture and structure information is repaired firstly. Meanwhile, this algorithm overcome the difficult problems of the inpainting algorithm based on texture synthesis searching the optimal matching block globally easily causes some matching errors. In addition, to further improve the effect of image inpainting, this algorithm assigns the size of repaired pixel block based on the local characteristics of image. Taking into account the image includes various types of repaired areas and single-layer inpainting algorithms can not be satisfied with the effect of inpainting, this paper presents a new image multi-level-inpainting algorithm based on partial differential equation and texture synthesis. This algorithm repairs image by using the inpainting algorithms based on partial differential equations in the first place and then deals with unsatisfied areas by using the inpainting algorithms based on texture synthesis. The experimental results show that the two kinds of algorithms in this paper improve the quality of image inpainting.

Related Dissertations

  1. Video Coding Research Based on Texture Characteristics,TP391.41
  2. Research on Methods of Medical Ultrasound Image Denoising,TP391.41
  3. Research on Restoration of Images Collected by Reconnaissance System in Near-Space,TP391.41
  4. Research on Learning-Based Low-Level Vision Problem,TP391.41
  5. Study on the Distribution Pattern and Priority Areas of Endangered Terrestrial Animals in Jiangsu Province,Q948.2
  6. The Research on Texture Synthesis Technology from Cloud Theory & Been Evolution Genetic Algorithm,TP391.41
  7. Tongyu County in the implementation of talent development strategy analysis,C964.2
  8. On the participation of civil execution distribution system,D925.1
  9. Selection of priority industrial development of the West Triangle Area,F127
  10. A Comparative Study on the Traditional Landscape Poetic Priority and Landscape Painting Image of Tang Dynasty and Song Dynasty,I207.22
  11. The Research of Medical Ultrasound Image Denosing Algorithms Based on Multiscale Geometric Analysis,TP391.41
  12. Research on Contourlet Transform and Its Application on Image Processing,TP391.41
  13. Texture Synthesis Algorithm Based on Samples,TP391.41
  14. The Study on the Role of Government to Perform Its Duties in Suzhou Public Transport Priority Development,U491.17
  15. Image Restoration Method Research and Its Application Based on Probability PCA,TP391.41
  16. Literal Interpretation of Administrative Litigation,D925.3
  17. Design and Simulated Implementation of Double-Link Failures Protection Algorithms in Traffic Grooming WDM Mesh Networks,TN929.1
  18. Research on Simulation and Scheduling Strategy in ARRAY Manufacturing System for TFT-LCD Fab,TN873.93
  19. Research and Realization of Image Denoising Based on Wavelet Transform,TP391.41
  20. Research on Simulations and Modified Algorithms of Median Filter,TP391.41
  21. Research on Adaptive Median Filtering Algorithm Applied in Image Processing,TP391.41

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net  Mobile