Accession Number : ADA254907


Title :   Application of Machine Learning Techniques for Effective Retrieval in Image Database


Descriptive Note : Final rept.,


Corporate Author : JACKSON STATE UNIV MS


Personal Author(s) : Gudivada, V N ; Raghavan, V V


Full Text : https://apps.dtic.mil/dtic/tr/fulltext/u2/a254907.pdf


Report Date : 30 Apr 1992


Pagination or Media Count : 14


Abstract : Recently, there has been widespread interest in various kinds of database management systems for managing formatted information. Depending upon the domain and the nature of formatted data, these systems are variously referred to as Multimedia Information Systems, Spatial Databases, Pictorial Information Systems, and Image Database Systems. However, the image data models employed in these systems are not based on any general framework. The model is rather extracted often from the implemented system and hence these data models are shaped by the idiosyncratic characteristics of the domains. Similar kinds of problems plague the query language design. By studying the application requirements and the limitations of the proposed approaches, we envision a multi-layered structure for retrieval. The various layers in the scheme are: Physical Layer, Spatial and Shape Layer, Iconic and Attribute Layer, and Conceptual Layer. These layers are not designed to operate in isolation but rather work in cooperation. To avoid redundancy in representation the layers are structured to form a lattice. The layers can also be viewed as multiple representations for the same object. This framework is expected to be highly flexible enough so that it can be useful across several application areas.


Descriptors :   *DATA BASES , *DATA MANAGEMENT , *FORMATS , REQUIREMENTS , MODELS , SHAPE , ISOLATION , LANGUAGE , COOPERATION , PLAGUES , APPROACH , REDUNDANCY , IMAGES , LIMITATIONS , STRUCTURES , LAYERS , INFORMATION SYSTEMS , MANAGEMENT


Subject Categories : Computer Systems


Distribution Statement : APPROVED FOR PUBLIC RELEASE