Abstract:An image stored in image database systems is assumed to be associated with some content-based meta-data about that image, that is, information about objects in the image and absolute/relative spatial relationships among them. An image query for such an image database system can generally be handled in two ways: exact picture matching and approximate picture matching. In this paper we show the intractability of matching of spatial relationships between a query image and an image stored in the database. In particular, our results suggest that one would not expect to have polynomial-time algorithms for finding the exact picture-matching and computing the maximal similarity between a query picture and a database picture, unless P = NP. 1. Introduction. Image database systems have been very active over the past 20 years. With the explosive interest for the last 10 years in multimedia systems, content-based image retrieval has attracted the attention of researchers across several disciplines, including computer vision, pattern recognition, human-computer interac- tion, and image databases. A recent survey on content-based image retrieval was given by Smeulders et al. (5). We (8, 9, 10, 11, 12) have proposed to develop a mathematical and algorithmic approach to modelling content-based image database systems. In this short paper, we intend to show the intractability of matching of spatial relationships between a query image and an image stored in the database. In partic- ular, one would not expect to have polynomial-time algorithms for finding the exact picture-matching and computing the maximal similarity between a query picture and a database picture, unless P = NP. 2. Specifying a General User Query. We first show how to specify a general user query, discussed in (12). A real picture is assumed to be associated with some content-based meta-data about that picture, that is, information about objects in the picture and absolute/rela- tive spatial relationships among them. An object in a real picture corresponds to a sig- nificant element of the image. Depending on the application, the significant elements of the image can be pixels, lines, regions, etc. A spatial relationship among objects is relative if it is determined by the position of the centroid of its objects. A spatial

A User-Dependent Definition of the Information in Images and Its Use in Information Retrieval

Image retrieval based on incremental subspace learning

Knowing a tree from the forest: art image retrieval using a society of profiles.

Modeling Image Data for Effective Indexing and Retrieval in Large General Image Databases.

Extending Databases to Support Image Editing

ON INTRACTABILITY OF SPATIAL RELATIONSHIPS IN CONTENT-BASED IMAGE DATABASE SYSTEMS

A General Approach to Indexing and Retrieval of Images in Image Databases

&Lt;title>information Embedding Based on User's Relevance Feedback for Image Retrieval</title>

GIS Semantics-Based Approach of Remote Sensing Image Retrieval

Information structures and uncertainty in an image information system

An Image-Semantic Ontological Framework For Large Image Databases

Relevance Feedback for Visual Data Retrieval

Using High Dimensional Indexes to Support Relevance Feedback Based Interactive Images Retrieval

Mutual Information calculation on different appearances

Contextual Visual Similarity

Why informatics and general science need a conjoint basic definition of information

Using Context Dependent Semantic Similarity to Browse Information Resources: an Application for the Industrial Design

Pattern retrieval in image databases based on adaptive signal decomposition

Using Information Content to Evaluate Semantic Similarity in a Taxonomy

Visual Ontology Construction for Digitized Art Image Retrieval

Image Retrieval Model Providing Semantics and Visual-Features-based Query for Users