Multimedia Databases: Performance Measure Benchmarking Model (PMBM) Framework

Abstract / Description

This paper presents the Performance Measure Benchmarking Model (PMBM) Framework for evaluating multimedia databases using Content-Based Image Retrieval (CBIR) techniques. The study addresses challenges associated with inconsistent performance measurement of image retrieval systems by developing a benchmarking framework based on image descriptors such as texture, color, and shape. Using De Groot's empirical research cycle, the framework was developed through image preparation, query definition, data collection, evaluation, and framework construction. Experiments conducted on CBIR image datasets assessed retrieval performance using accuracy, precision, recall, and F-measure metrics. A Java-based software evaluation tool was subsequently developed to validate the effectiveness of the PMBM framework in qualifying and quantifying multimedia database performance.

Keywords

Suggested Citation