원문정보
초록
영어
In Digital image processing; many researches have been done on image denoising so far. Nowadays, the noise detection from an image is the most challenging task. Though, the various algorithms introduced for the detection of noise type from a noisy image, but these algorithms work only for detection of single type of noise. To overcome the limitation of the previous built algorithms, we investigate the data mining technique called Support Vector Machine. The SVM is a powerful supervised learning method which is to be used for the detection of mixed noise models. Broadly, this technique detects the different types of noise from a mixed noise image; noise can be either single or mixed type of noise. The different parameters have combined to describe the properties of these different noise models so as to perform the detection. The detecting algorithm has been achieved by applying the SVM on the training dataset of different medical images and further extensive tests are performed on the test dataset for detection of each noise type model. This detection technique clearly outperforms various techniques with the high accuracy of results for different proposed noise models.
목차
1. Introduction
2. Proposed Noise Models
3. Support Vector Machines
4. Multiclass Support Vector Machine
5. Methodology Of Proposed Technique
5.1 . Data Collection
5.2. Preprocessing
5.3. Feature Selection and Extraction
5.4. Train the SVM
5.5. Test the SVM
5.6. Evaluate
6. Experimental Analysis
6.1. Contrast
6.2. Correlation
6.3. Energy
6.4. Homogeneity
6.5. Mean
6.6. Variance
6.7. Standard deviation
6.8. Entropy
6.9. Smoothness
6.10. Kurtosis
6.11. Skewness
6.12. IDM (Inverse Difference Moment)
7. Results and Discussions
8. Conclusion
References