This chapter describes a novel unsupervised information-theoretic adaptive filter (UINTA) for image restoration [6,9]. UINTA restores pixels by comparing pixel values with other pixels in the image that have similar neighborhoods. The underlying formulation relies on an information-theoretic measure of goodness combined with a nonparametric model of image statistics. UINTA minimizes a penalty function that captures the entropy of the patterns of intensities in image regions. UINTA filtering, obtained as the derivation of the entropy, is nonlinear. UINTA operates without a priori knowledge of the geometric or statistical structure of the signal, but relies instead on some general observations about the entropy of natural images. It does not rely on labeled examples to shape its output, and is therefore unsupervised. UINTA automatically learns the true image statistics from the degraded input data and constructs a filtering strategy based on that model, making it adaptive. Moreover, UINTA adjusts virtually all its important internal parameters automatically using a data-driven approach and information-theoretic metrics. Because UINTA is nonlinear, nonparametric, adaptive, and unsupervised, it can restore a wide spectrum of images with very little parameter tuning.