TOYON RESEARCH CORPORATION — Department of Defense STTR Phase I: AF15-AT35

Description

ABSTRACT: Toyon Research Corp. and the Penn State Univ. propose research and development of innovative algorithms for classifying objects and activities observed in high-dimensional data, including video and hyperspectral imagery. The proposed algorithms include novel feature clustering techniques to enable effective characterization of intra-class and inter-class appearance variations in datasets containing a small number of labeled, and a large number of unlabeled, high-dimensional feature vectors. The proposed development is expected to provide significant improvements in object and activity classification performance, including maximization of the probability of correct classification and minimization of false declaration rates for real-world applications including highly variable clutter and object and activity types not represented in the labeled training data. The proposed algorithmic framework is of a general nature and utility, and will be demonstrated using multiple real-world image and video datasets in Phase I. In Phase II, real-time prototype software will be developed and demonstrated for additional real-world applications, and integration in AFRL systems will be supported.; BENEFIT: The successful completion of this research will result in the development of technology capable of monitoring data from large numbers of disparate imaging and video sensors, with automated or semi-automated recognition and extraction of objects and activities of interest. Work in this proposed effort has the potential to move image/video processing beyond baseline machine vision towards intelligent vision. DoD applications include identification of militarily relevant objects in cluttered scenes containing large numbers of distractor objects, and recognition of threatening activities in the midst of benign activities. Homeland Security and TSA could use this technology to identify potential terrorist threats. Consumer-focused manufactures could develop a new generation of products that interact with and intelligently assist humans in performing a wide range of tasks.