Efficient Compact Bio-inspired Sensory Information Processing

We propose to develop enhanced IR and intensified visible senors using a novel image processing system based on processing mechanisms and signal representation derived from the study of mammalian retinas. In a set of stacked strata that span depth of theinner plexiform layer, the the mammalian retina forms about a dozen different images representing spatial-temporal feature sets derived from input stimuli. These representations are carried intact to higher visual centers by different populations of opticnerve fibers and contain all the information we rely upon when interpreting the visual world. These parallel feture sets comprise the native representation of visual information, a kind of native visual language. Using this language, we can formulaterobust solutions to essential image processing problems such as image restoration, enhancement, and analysis, motion detection, feature extraction, and identification. Through this continuing effort, we will refine our quantitative understanding of boththe mechanisms and function of retinal image processing circuitry and will implement in real-time imaging devices the type of space-time processing that underlies our own ability of understanding and interacting with the visual world. The performance ofthe enhanced imaging systems will bedemonstrated and characterized in field tests under a variety of demanding environmental and signal conditions. The proposed research and development effort aims to enhance current capabilities in sensory informationprocessing applications by extending sensor operating ranges, resilience against sensor deficiencies via post processing, providing robust operation in noisy or cluttered environments, and enhancing image delivery and display technologies. At the sametime, results from this effort are expected to extend our understanding of biological vision and to provide modeling and experimental tools for further studies. Applications of Navy interest include automatic target detection and tracking, night vision,real-time image analysis and recognition, as well as image compression. New or enhanced military products that may result from this effort include vision systems for targeting and guidance systems, surveillance and reconnaissance systems, night vision,battlefield communication systems, smart munitions and sensor payloads. Commercial applications of the proposed research and development include civilian aviation both in the form of landing instruments for commercial aircraft and as ground-basedinstrumentation for flight control and runway surveillance, industrial inspection and process control, traffic monitoring, commercial asset protection, and visual communications.