Image Retrieval
Query by Example: finding more images in a
collection that are like the target.
Image retrieval comprises any technology designed to locate particular images within large collections, such as those found on the world wide web. Although many techniques exist, including some that look at the text and labels surrounding an image, I am most interested in content-based image retrieval, which seeks to retrieve relevant images based on an automatic analysis of their contents.
The heart of a content-based image retrieval algorithm is its formula for measuring the similarity between any pair of images. I have developed a technique that compares images based upon a combination of the color, texture, and arrangement of entities within the image frame.
A Closer Look at Boosted Image Retrieval, N. Howe. International Conference on Image and Video Retrieval, July 2003. [PDF] [PS.GZ] [BibTeX] [PowerPoint].
Analysis and Representations for Automatic Comparison, Classification, and Retrieval of Digital Images , N. Howe. Ph.D. Thesis, Cornell University, May 2001. [PDF] [PS.GZ] [BibTeX] [PowerPoint].
Data as Ensembles of Records: Representation and Comparison, N. Howe. Proceedings of the Seventeenth International Conference on Machine Learning, 2000. [PDF] [PS.GZ] [BibTeX] [PowerPoint].
Integrating Color, Texture, and Geometry for Image Retrieval, N. Howe and D. Huttenlocher. Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2000. (1) [PDF] [PS.GZ] [BibTeX].
Using Artificial Queries to Evaluate Image Retrieval, N. Howe. IEEE Workshop on Content-Based Access of Image and Video Databases, 2000. (2) [PDF] [PS.GZ] [BibTeX] [PowerPoint].