A heterogeneous processing (HP) framework for multimedia query processing

Defines a novel system based on a hierarchical computational and data abstraction model to handle multimedia queries involving face recognition, scene matching, and speech recognition. For this purpose, the system integrates the state-of-the-art research in diverse areas of the computational sciences, including parallel processing, computer vision and image processing, artificial intelligence, multimedia databases, and speech and signal processing. This system uses a heterogeneous processing framework consisting state-of-the-art image understanding and signal processing algorithms in addition to content-based retrieval techniques. These algorithms comprise of parallel coarse-grained (MIMD) symbolic and fine-grained (SIMD) numeric algorithms at various levels of this hierarchical model