This was a class project designed by Professor Min Chen. The goal of this project was to make a very simple search engine for images, using no textual meta data whatsoever. Later stages of the project added user interaction in the form of Relevance Feedback, which was a simplified form of what is described in [Rui et al. 1998]. I also read another article for a broader perspective, but only ended up using techniques described in the classroom lectures [Müller et al. 2000].
For the convience of the end user, image features are saved in a raw format using .NET binary file serialization; however, these features are always normalized prior to any statisical analysis or image retrieval.
Müller, H., Müller, W., Marchand-Maillet, S., Pun, T. and Squire, D.M., 2000. Strategies for Positive and Negative Relevance Feedback in Image Retrieval. In IN PROCEEDINGS OF THE 15TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION (ICPR 2000. IEEE, pp. 1043–1046.
Rui, Y., Huang, T.S., Ortega, M. and Mehrotra, S., 1998. Relevance Feedback: A Power Tool for Interactive Content-Based Image Retrieval,