Detecting and Extracting Image Similarities, Differences and Target Patterns

MTL Systems, Inc. and Dr. Francis Quek of Wright State University propose a unique and commercially-viable solution to the problem of automatically detecting, extracting, and recognizing changes and similarities in target patterns in sequences of images.Our proposed Optimized (image) DIscriminant Nexus** (ODIN) program will synergistically integrate several methods, including (1) a demonstrated, revolutionary difference detection, and pattern recognition technique called TWIST* (TWo-axis Image SortingTechnique), (2) piecewise (image) rigidity concepts, (3) Hough Transforms, and (4) digital filtering. ODIN captures image complexity information from a sequence of video frames, into a concise feature set, to provide a reliable, accurate, robust method todetect changes or similarities, or to recognize target patterns in multi-modal (EO, IR, SAR, etc.) digital imagery. In Phase I, MTL will perform (1) a prototype development, (2) experimental and analytical feasibility assessments using realistic use-caseimages, (3) a prototype demonstration, (4) a preliminary system design to carry forward into Phase II, and (5) an initial assessment of commercialization potential.*Patent pending** NEXUS = A connected group of discriminant techniques The ability to automatically and accurately recognize and track changes, similarities, or object patterns and their locations in a surrounding environment, independent of the particular environment features, is of great importance to government andcommercial enterprises alike. A particularly compelling commercial opportunity is a video security application, using ODIN techniques to recognize vehicles or individuals of interest based on their static and dynamic behavior. Commercial developers andmarketers of video-based security systems are a ready market for ODIN. The ODIN process is much more cost-effective for commercial and military operations than current products, since it operates faster, and requires less data storage space (for images ormodels) than other techniques. ODIN can replace slow and user interaction-demanding systems with a low-cost, automated alternative.