Dissertation > Excellent graduate degree dissertation topics show

Keep the texture image restoration algorithm

Author: HuangXiaoJun
Tutor: LiuXiaoYun
School: University of Electronic Science and Technology
Course: Control Theory and Control Engineering
Keywords: Image Restoration Remove noise Deblurring non-local means filter Texture maintain
CLC: TP391.41
Type: Master's thesis
Year: 2011
Downloads: 40
Quote: 0
Read: Download Dissertation

Abstract


Because many factors, image capture and transfer process usually degraded, its typical performance: vague and noise. The purpose of image restoration is to improve to fixed degrade image quality, and as much as possible to restore the original image. Deblurring and remove noise are two common types of image restoration problems. Many outstanding image restoration algorithm estimate is very close to the original non-degraded image processing of image detail lacking. There are always many edges and points constitute the details in the actual image, and most of the restoration algorithm solution is too smooth, resulting in the loss of texture detail loss means a loss of texture information. In order to maintain a good image texture while noise removal and deblurring, do the following main tasks: (1) remove the noise process to maintain the image texture for noise image recovery problem, based on non-local means filter (NL-means filter) has put forward an improved algorithm. The basic idea is to replace the gray value of each of the corresponding pixels using a weighted average of the gray values ??of all pixels in the degraded image. The improved algorithm is not only used in the calculation of the weights of the similarity of the characteristics of the original algorithm compares neighborhood, and also take advantage of the spatial distribution characteristics of the gray values ??of the pixels in the image. By a large number of simulation experiments to verify the improved algorithm can basically keep the live image texture and edge, better than the original NL-means algorithm and the original NL-means algorithm is a special case of the improved algorithm. (2) for the blurred image deblurring process to keep the recovery image texture and design a way to keep the image texture deblurring algorithm (Texture-Preserving Image Deblurring a TPID). The algorithm deblurring process is divided into two steps, first using the Wiener filter to recover the blurred image to get a useful signal components to achieve the minimum loss Noisy results, then use the improved NL-means algorithm to suppress the leak colored noise. The algorithm combines the Wiener filter to remove blurring fast, effective and improved NL-means algorithm suppress strong noise, keep the texture of the advantages of one, so that the restored image is largely intact to maintain the texture. Comparative tests with other outstanding deblurring algorithm, subjective visual evaluation and objective numerical evaluation of the proposed algorithm is better than other methods in maintaining the image texture. In order to improve the execution speed of the algorithm, we achieve parallel CUDA (Compute Unified Device Architecture) programming model based on GPU (Graphic Process Unit) of the algorithm. The experimental results show that the execution speed can basically meet the demand for practical application.

Related Dissertations

  1. Research on Restoration of Images Collected by Reconnaissance System in Near-Space,TP391.41
  2. Research on Learning-Based Low-Level Vision Problem,TP391.41
  3. Image Restoration Method Research and Its Application Based on Probability PCA,TP391.41
  4. The Research of Image Restoration Based on Phase Diversity Method,TP391.41
  5. Based on public safety characteristics of millimeter-wave radiation image,TP391.41
  6. Exercise and defocus blurred image restoration,TP391.41
  7. Image restoration of key technologies,TP391.41
  8. Remote sensing image reconstruction algorithm oriented IICCD camera is not completely random sampling,TP751
  9. A Research of Image Denoising and Restoration Based on Total Variation Method,TP391.41
  10. Research on Digital Image Inpainting,TP391.41
  11. Research of Underwater Image Restoration Algorithm Based on Backward Scattering Noise Model,TP391.41
  12. Common types blurred image restoration Research and Implementation,TP391.41
  13. Medical endoscopic imaging systems and image restoration research,TP391.41
  14. Research and Application of the Panoramic Annular Lens Imaging System,TP391.41
  15. Research of Digital Image Inpainting Technology,TP391.41
  16. Weak signal detection mechanism of stochastic resonance network model and its application,TN911.23
  17. Study on the Algorithms of Adaptive Optical Images Restoration,TP391.41
  18. SAR Image Despeckling Based on Nonlocal Means Filtering,TN957.52
  19. Footage plaque damage repair technology research,TP391.41
  20. Image Motion Deblurring,TP391.41
  21. Research of Single Image Motion Deblurring and Video Sequence Stabilization,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