CAREER: Coherent Computational Imaging: Micro Measurements in a Macro World

Why bridge microscopy and photography? In microscopy there is a degree of freedom that is just not available in photography. When scenes are microscopic, a camera lens can be placed very close. This has a fundamental impact on the level of details that can be captured. When the lens aperture is large relative to the distance to the scene, the effects of diffraction are at their smallest. However, in photographic settings, we are typically interested in capturing images remotely; we want to be able to stand back from the scene and make observations at a distance. This is particularly important in computer vision applications such as surveillance and remote sensing, where often we want the camera to be so far away as to be invisible to the naked eye. When we view objects from a distance, diffraction limits the level of detail we can observe. For instance, consider the problem of detecting a face from a photograph taken at a distance of 1km away. If the photograph is taken with a typical SLR camera lens with aperture size around 20mm, the diffraction blur will be on the order of 100mm nearly the full size of the face. Note that this is independent of the focal length or magnification. There is no way to optically zoom in to capture a higher level of detail. The only way to decrease the diffraction blur is to use a lens with a much larger aperture. While lenses as large as 300mm diameter can be purchased commercially, they are extremely bulky and come at an exorbitant cost, prohibiting use in wide array of applications. Ultimately, we face severe practical limitations on how large the aperture can be, forcing us to seek solutions by other means. In this proposal, we focus on the problem of capturing photographs where objects in the scene are far from the aperture. Far can be anything from several meters to kilometers. Our goal is to increase the precision of cameras in three key areas: image, depth, and material acquisition. To achieve this goal, I will build on my extensive research in computational imaging. I will draw from current microscopy techniques to look for methods to increase measurement precision for macroscopic scenes.