Dissertation > Excellent graduate degree dissertation topics show

Research on the Methods of Blind Image Restoration

Author: WangZuoNa
Tutor: GuoYongCai
School: Chongqing University
Course: Optical Engineering
Keywords: Blind image restoration NAS-RIF algorithm Wavelet Transform Regularization
CLC: TP391.41
Type: Master's thesis
Year: 2008
Downloads: 271
Quote: 0
Read: Download Dissertation

Abstract


The image restoration techniques are widely used in aerial mapping, remote sensing, astronomy, medicine, criminal reconnaissance. Many existing recovery methods are carried out under the premise of fuzzy operators know exactly, this image restoration is known as the classic image restoration. More common fuzzy unknown or uncertain process, which requires information drawn in some way from the degradation of image degradation, such using only some or very little degradation system information, it is estimated that the original image is known as the blind image restoration. Blind image restoration does not depend on the transfer function of the system, and thus has a wider range of practical value. This article is around blind image restoration method study. Thesis elaborated the basic method of blind image restoration, image degradation model and image restoration process ill-posed problem. The original non-negative support domain recursive inverse filtering (NAS-RIF) algorithm based on a blind image restoration algorithm based on spatially adaptive regularization techniques improved. The algorithm is introduced in the cost function of the original NAS-RIF algorithm two spatially adaptive weighted items, are used to ensure a realistic and smooth image restoration, adaptive weighted items to be based on the local characteristics of the observed image and the noise variance obtained. Add the regularization term, in order to achieve the purpose of noise suppression. A detailed analysis of the regularization operator the sub selection method, high pass low resistance operator should be selected as the regularization operator. In Least Squares recovery based on using different regularization parameter to restore the blurred image, regularization parameter should be based on the local characteristics of the image and noise variance to select the conclusions drawn by comparison recovery result. Proposed a method to estimate the noise variance according to the observation image, and thus do not need to know a the noise variance priori conditions. On the basis of the estimated noise variance, the proposed method based on the estimated local variance of the noise and image to select the regularization parameter. In the solution using the conjugate gradient algorithm to solve. The simulation experiments on three different background and different signal-to-noise ratio of the image. Respectively to improve the signal-to-noise ratio gain improved algorithm (△ SNR) than the original algorithm: 0.2073db, 1.0239db, 2.8628db. Improved algorithm to obtain a better image restoration effect. NAS-RIF algorithm extended to the wavelet domain, an adaptive regularization method with the NAS-RIF blind image restoration algorithm for wavelet domain algorithm combined. Degraded image wavelet decomposition image information in different sub-bands. The frequency and direction of the image characteristics for each of the sub-band, respectively, the introduction of different regularization constraint items. Estimated in each sub-band noise variance, and propose a method to select the regularization parameter according to the noise variance and image local variance. The experimental results show that: the proposed algorithm, the restoration effect relative spatially adaptive regularization method has improved to some extent.

Related Dissertations

  1. Feature Extraction, Selection and Combination in Lipreading,TP391.41
  2. Research on Identification System of Cashmere and Wool Fiber,TS101.921
  3. Characteristics of sensory stimulation evoked,R318.0
  4. Network transmission ROI image coding algorithm,TN919.81
  5. Image Fusion Algorithms Based on Multi-scale Analysis,TP391.41
  6. Feature Extraction Technologies Research and Implementation of 3D Models Based on Wavelet Transform,TP391.41
  7. An Algorithm on Clustering and Anomaly Detection for Multiple Data Streams,TP311.13
  8. Electrical Impedance Tomography numerical solution of the problem in some studies,O441.4
  9. Traffic Prediction Model Based on Wavelet and Markov,TP393.06
  10. Inspection and Recognition on a Class of Tool Processing Trace,TP391.41
  11. The Realization and Study of a Scalable Data Encoder Based on SPIHT,TN762
  12. Based on non- parametric statistical characteristic quantities Gaussian kernel network traffic anomaly detection method,TP393.07
  13. Radial Shearing Interferometry Test Technology,TH744.3
  14. Iris feature -based digital watermarking algorithm,TP309.7
  15. Helicopter main gearbox fault diagnosis and failure prediction technology research,V267
  16. Of speckle noise removal in medical ultrasound image,TP391.41
  17. Comprehensive Research on Intellective Deformation Analysis Method,P209
  18. Learning Rates of Least-square Regularized Regression with Polynomial Kernel on a n-dimensional Simplex,O174.41
  19. ECG Denoising and QRS detection algorithm,TH772.2
  20. Research on Dodging Methods of Optical Remotely Sensed Image,TP751
  21. Denoising Research of Underwater Images Backscatter Based on the Theory of Wavelet,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