This Presidential Young Investigator Award is for support of John Daugman at Harvard University. His research deals with information processing in both biological and computational vision. Focusing on early mechanisms of visual signal processing, he has used image-processing methods to simulate the basis functions in neural laminae which extract two-dimensional image structure. The inferred functions can be modeled as oriented, multiscale, two-dimensional wave packets that simultaneously and optimally resolve structural and positional information. Daugman now is developing a machine-vision processor that will decompose images, at video rates, into oriented structural primitives using parameters based on neurobiological recordings of the properties of cells in the visual cortex of the brains of primates. The significance of this work is that it is likely to lead to a better understanding of natural vision systems -- how they extract information from images and represent it, and how they infer the properties of objects.
