QUIMaS (Queen's University Image Management System): a museum photographic database

Abstract

This paper reports on the construction of an interactive image management system. The system's operation is improved by image processing techniques and parallel processing technology. Although image processing is largely a well understood problem, the technology for supporting it is still immature. Basically, it is a highly computationally intensive activity, and the purpose of our project was to address this very problem using parallel processing technology in the form of transputers. More specifically, we will describe the use of a transputer network in the development of an image database consisting of a collection of historical museum photographs. During the first phase of the project, we developed a prototype system consisting of a small number of photographs using a simple hardware configuration. It made use of the hard disk of the PC to store the photographs and was implemented using a single transputer. During the second phase, the full hardware configuration was implemented, including the building up of the database of around 1,000 photographs. The final system consists of an IBM-compatible PC equipped with the necessary image processing and mass storage facilities (650 Mb hard disk) and linked to a network of six transputers. The user interface is provided through a serial mouse and a keyword menu. Users are able to retrieve images in three different modes: (i) a predefined keyword list, which is arranged in a hierarchical fashion to select images relating to a particular subject or geographical area; (ii) the Browse mode allows the user to display a set of up to twelve photographs in miniature ‘stub’ format at one go (this is similar to the turning of the pages of an album). Individual items can then be selected and displayed in full size, if necessary; (iii) if the museum reference number is known, the user can type in the number to retrieve the relevant item. Section 1 will provide a brief review of the background to the project and the use of transputer technology in image processing. The selection and documentation of the photographs are described in Section 2. Section 3 describes the hardware and software configuration of the single-transputer-based prototype, and the full image database system implemented on a network of transputers. The organisation of the database, both textual and image data, is explained in Section 4. This is followed by a description of the processing techniques for image enhancement and compression, and a discussion of the results of both compression and decompression processes. The final section will discuss the menu-based, mouse-driven user interface, including the various search facilities and display modes used to view images from the database.