AFRL Demonstrates Biologically Inspired Foveal Sensing in Infrared Cameras Published March 13, 2008 By Paul McCarley AFRL/RW EGLIN AIR FORCE BASE, Fla. -- Under an AFRL Small Business Innovation Research contract, Nova Sensors developed a family of camera systems providing variable-acuity superpixel imaging (VASI™) capabilities. This advanced imaging technology permits maximum-resolution focus on regions of interest from the visible to IR spectrum, while significantly reducing the bandwidth required for image readout and processing. The VASI-based cameras' built-in capacity for frame-to-frame, dynamically programmable spatial resolution emulates the biological paradigm of foveal vision. The biological makeup of the human eye entails a small area of high resolution (the foveal region), with radially decreasing peripheral clarity. VASI technology mimics the human foveal vision system's function, in that (after performing on-chip spatial binning), system sensors sample the "most important" image information at high spatial resolution and the "less interesting" information at lower spatial resolution. This guarantees that key image content is preserved. It also ensures both that the total field of view (FOV) is always monitored (should other interesting objects appear in the periphery) and that the highest possible frame rates are produced for a fixed bandwidth or, alternately, for the lowest possible bandwidth. Thus, even as VASI sensors are capturing and retaining essential image content, they are reducing the amount of image data received from large-format focal plane arrays (FPA). An integrated-circuit architecture enables VASI readout of the entire FOV at high frame rates. The integrated circuitry facilitates this coverage by permitting larger pixels--superpixels--to be dynamically formed on the FPA in regions of relative unimportance, thus reducing the total number of pixel values that must be multiplexed off the FPA. In addition, this architecture can accommodate multiple high-resolution foveal regions "flown" around the imager's FOV at a frame rate that facilitates sampling of critical targets at the highest possible spatial resolution the imager can produce. The applications for multiresolution sensor (i.e., "foveal") architecture are plentiful. One such use of this biologically inspired technology lies in multiple-target tracking, wherein target characteristics such as vehicle motion demand incorporation of representative foveae in order to generate sequential imagery that reflects not only high-frame-rate operation but also the required level of information recovery for mobile objects of interest.