Transactions on Machine Learning and Data Mining (ISSN: 1865-6781)
Volume 2 - Number 1 - July 2009 - Page 3-22
Content-Based Image Indexing and Retrieval in an Image Database for Technical Domains
M. Ewald
Institute of Computer Vision and Applied Computer Sciences, Leipzig, Germany
Abstract
The availability of a variety of sophisticated data acquisition instruments has resulted in large repositories of imagery data in different applications like non-destructive testing, technical drawing, medicine, museums and so one. Effective extraction of visual features and contents is needed to provide meaningful index of and access to visual data. In this paper, we propose an image database architecture which can be used for most industrial problems. The image database is able to handle structural representation of images. Indexing is possible object based, spatial relation based, and by a combination of both. The query can be a textual query or a image content based query. We propose how the image query is processed, how similarity based retrieval is performed over images and how the image database is organized. Results are presented based on an application of ultra sonic images from non-destructive testing.
Download Paper
(437 KB)