A Local Structure Matching Approach for Large Image Database Retrieval
The combination of a local structure based shape representation and a histogram based indexing structure is proposed to fast localize objects from large database. Four novel geometric attributes are extracted from each local structure. They are invariant to translation, scaling, rotation and robust to adverse distortions and noise. The search space is pruned by means of histogram intersection and the computation cost of the query is linear to the number of input features. The matching is performed by a non-metric similarity measure with regard to significance in reconstruction of query image and discrimination of different models. The concepts proposed were tested on thousands of images. The result manifests its efficiency and effectiveness.
KeywordsLocal Structure Indexing Structure Query Image Zernike Moment Shape Match
Unable to display preview. Download preview PDF.
- 7.Mokhtarian, F., Mackworth, A.K.: The Curvature Scale Space Representation: Theory, Applications and MPEG-7 Standardization. Kluwer Academic Publishers, Dordrecht (2002)Google Scholar
- 8.Mehrota, R., Gary, J.E.: Similar-shape retrieval in shape data management. IEEE Computer 28, 57–62 (1995)Google Scholar
- 12.Koffka, K.: Principles of Gestalt Psychology. Harcourt, Brace and Company, New York (1935)Google Scholar
- 13.Fisher, M., Smith-Gratto, K.: Gestalt theory: a foundation for instructional screen design. Journal of educational technology systems 27(4), 361–371 (1998-1999)Google Scholar
- 14.Chang, D., Dooley, L., Tuovinen, J.E.: Gestalt Theory in Visual Screen Design-A New Look at an Old Subject. In: The Seventh World Conference on Computers in Education, Copenhagen, Denmark (2001)Google Scholar
- 21.Brady, M.: Criteria for representations of shape. In: Beck, J., Hope, B., Rosenfeld, A. (eds.) Human and Machine Vision, pp. 39–84. Academic Press, New York (1983)Google Scholar