TOYON RESEARCH CORPORATION — Department of Defense SBIR Phase I: SOCOM163-008

Description

Toyon Research Corp. proposes a feasibility study in preparation for development of an innovative intelligence repository which extends and scales current database capabilities to the paradigm of wide-area, multi-modal, millimeter-accuracy reconnaissance. The envisioned software architecture is based on a fusion of traditional key-based indexing with both geographic and content-based indexing schemes, which enables the fast search, processing, and dissemination of information in multi-modal datasets. At the lowest level, data is organized into tree-like computation graphs, whose structure can be adaptively tuned to increase performance of data correlation and information processing tasks, such as wide-area search and classication of multi-modal imagery. Hierarchical and graph-based representations also make processing pipelines, such as geo-registration, 3D-reconstruction, and CAD-model analysis, intuitive and simple by providing a unied virtual-le representation. The resulting specication should be highly tunable and adaptable to various data modalities (e.g. text, multi-modal imagery, point-clouds), allowing developers to adopt and extend a system to match their use-case without redening the schema. Furthermore, the system must be capable of real-time performance, with extensions employing massively distributed and decentralized network architectures.