We propose to extend our current lead in the area of deep astronomical imaging to decipher the secrets of galaxy formation and evolution. In particular, we will prepare for the imaging from the Large Synoptic Survey Telescope (LSST) to which we will soon have privileged access. The LSST project is truly a 'Big Data' project in astrophysics, which will produce 60 petabytes of imaging data at a rate of 20 TB per night, and which will initiate a profound chance in our profession, from gathering specific data to answer a question to data-driven exploration and discovery. We will develop algorithms to deal, in an automatic fashion, with two of the most important systematic problems affecting deep astronomical imaging: scattered light and foreground emission from Galactic cirrus. We know how to deal with this on scales of one or a few images, but will now include machine-learning techniques to automate the process, and make it work on huge datasets.
