Dissertation > Excellent graduate degree dissertation topics show

Based on kernel regression and non- local method of image denoising

Author: SongLi
Tutor: CaoHanQiang
School: Huazhong University of Science and Technology
Course: Communication and Information System
Keywords: Non- local approach Kernel function Kernel Regression Singular Value Decomposition Image Denoising
CLC: TP391.41
Type: Master's thesis
Year: 2011
Downloads: 80
Quote: 0
Read: Download Dissertation

Abstract


Kernel Regression analysis is a traditional regression analysis on the latest developments in image denoising , data mining, super-resolution analysis and other fields has been widely used. Non- local method is currently the most popular image processing method , which calculates the weight characteristics to consider the overall image , the process is much better than other methods , the disadvantage is large amount of calculation . In this paper, the National Science Foundation of Natural background , combined with the locality of kernel regression algorithm with non local algorithm overall, will be applied to image denoising , got better treatment effect . This paper first introduces the basic concepts of image processing as well as common noise model , recalling the traditional image denoising method to analyze the deficiencies of the traditional denoising methods , these methods are based on a specific model structure , since the model has been set, which limits their applications, kernel regression method a good improvement of the previous denoising these shortcomings , the disadvantage is not taken into account in the de-noising gray image information , this article by introducing local direction information to improve the classical kernel regression . Kernel regression weights in the calculation of analyzing only the pixels adjacent areas , in fact, distant region may also have some similar characteristics , so the kernel regression method has some limitations. This paper describes a non- local denoising good use of the image of global features , for a similar texture image , the method of the treatment effect is much better than other methods. However, its use is locally constant model , it is the kernel regression is equal to 0 in the order of a special case , by combining the high-level kernel regression resistance, can be a good way to improve the lack of non- local . Finally, experiments and analysis show that this method as compared with the conventional method , since a good combination of the local image features and global feature , making the image denoising better able to keep the texture and structure of the information, with better denoising effect .

Related Dissertations

  1. Modulation Pattern Recognition and Parameter Estimation for Ground Wave Radiation Sources,TN957.51
  2. Research on Methods of Medical Ultrasound Image Denoising,TP391.41
  3. Process Support Vector Machine and Its Application to Satellite Thermal Equilibrium Temperature Prediction,TP183
  4. Research on Contourlet Transform and Its Application on Image Processing,TP391.41
  5. Wavelets-based Signal Sparse Representation and Its Application in Image Denoising,TP391.41
  6. Astudy on Hierarchical-Matrix-Based Algorithm for Solving Electromagnetic Problems of the Volume Integral Equation,O175.5
  7. Research on Approaches of the Subjective Automated Assessment,TP391.1
  8. Capsule endoscopy and endoscopic image portable receiver system bleeding Recognition Algorithm,TP391.41
  9. Web news hot discovery system design and implementation,TP393.09
  10. Tetrolet sparse regularization learning with the sample image super-resolution algorithm,TP391.41
  11. Subspace-based analysis of Narrowband Interference Suppression in Spread Spectrum Communications,TN914.42
  12. Combination of multi-core based on the generalized Gaussian image classification method,TP391.41
  13. Kernel-based SOM time series prediction,O211.61
  14. Denoising problem solving variable weight fixed point algorithm,O177.91
  15. Research of Protein-Protein Interaction Extraction Based on Rich Feature and Multiple Kernels Learning,Q51
  16. Signal Processing Method Based on Singular Value Decomposition and Its Application to Mechanical Fault Diagnosis,TH165.3
  17. Research on Subspace Analysis Based Face Recognition Algorithm,TP391.41
  18. Research on Algorithm of Image Super-resolution Reconstruction,TP391.41
  19. SVD -based digital watermarking algorithm,TP309.7
  20. Research of Digital Watermarking Technique in Transform Domain,TP309.7
  21. Seismic data random interference suppression technology research,P631.44

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