A hierarchical approach for content-based echocardiogram video indexing and retrieval

Abstract

Content-based video retrieval is one of the most challenging areas of research in information sciences. Here we present a new method for efficient indexing, storage and retrieval of echocardiogram video data. The architecture and implementation details of the system have been discussed. A hierarchical state based indexing scheme has been used for faster retrieval of information related to states of objects present in echo video. In video segmentation phase, views as well as states and sub-states of objects in a video are identified. Accordingly, schemas are populated. An object-relational database approach has been adopted to store object information into a backend RDBMS. The proposed video query language can be used to compose queries based on video contents using annotations, states, events, spatial or temporal relations. For ease of specifying query, a user friendly interface has been provided. Retrieved segments can be played back one at a time in a separate window.