Content-based image retrieval by means of scale-space top-points and differential invariants

E. Balmachnova, B. Platel, L.M.J. Florack, B.M. Haar Romenij, ter

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review


The paper presents an approach to the automated image segmentationproblem for images of the digestive tract, using the color set back-projectionalgorithm. To implement this algorithm the image is transformed from RGBcolor space to HSV color space and quantized to 166 colors. At the end of thisprocess, the color set of the image is obtained and used in color regiondetection. The resulting color regions are then used in a content-based regionquery process. Experiments were made on a database with 960 color imagesthat represented: polyps, colitis, ulcer, ulcerous tumor and esophagitis, resultingin about 5,000 color regions. Intial query results have been satisfactory, and thedeveloped software tool is now used for medical teaching. This small imagedatabase used in this work is being extended to a larger set for the evaluation ofthese algorithms.
Original languageEnglish
Title of host publicationProceedings of the MICCAI 2007 workshop on Content-based Image Retrieval for Biomedical Image Archives: Achievements, Problems and Prospects, 29 October 2007, Brisbane, Australia
Place of PublicationAustralia, Brisbane
Publication statusPublished - 2007


Dive into the research topics of 'Content-based image retrieval by means of scale-space top-points and differential invariants'. Together they form a unique fingerprint.

Cite this