03 April 2010

Report: Database for Archiving Paintings (2007)


Title: Spectral Imaging Database by PCA & NMF for Archiving Paintings Application
Author: Yifan Zhou
Year: 2007
Pages: 72
Type of document: Internship report
Download: http://spectral.joensuu.fi/... (pdf, 3.67 Mb)


Table of Contents (short version):

1. Acknowlegments

2. Objective

3. Introduction
3.1. Advantages of using digital image processing techniques in archiving paintings application.
3.2. Reasons why we need to use spectral images instead of RGB images.
3.3. How to characterize spectral images to index and search in a database?

4. Capturing of spectral Images
4.1. Basic conceptions of acquiring spectral images.
4.2. Accomplished international projects.
4.3. One of the capturing systems in color research group.

5. Methods of PCA & NMF
5.1. Principles of pca and nmf
5.2. Adoption to spectral images transformation
5.3. Advantages and disadvantages

6. Index and Search by PCA and NMF in the Spectral Imaging Database
6.1. Construction of an imaging database
6.2. Conventional ways to index and search in an imaging database
6.3. My procedure to index and search in a spectral imaging database

7. Conclusions
7.1. Related work
7.2. Discussion of experiment results
7.3. Future research directions
7.4. Conclusion of this internship

8. References

9. Appendices
9.1. Spectral images in the spectral imaging database
9.2. Eigenvectors and non-negative vectors calculated by the whole sub-database
9.3. Index spectral images by PCA and NMF
9.4. Distance between selected V and original V, selected W and original W of 1% to 20% average selections by spectral metric of PCA
9.5. Accuracy of vectors vs. selections
9.6. Feature vectors
9.7. GFC and PSNR distance of random selections of 1% to 99%
9.8. List of functions
9.9. Abbreviation


In case you would like to share this publication, please acknowledge this source.

No comments: