Photography is just now in the early stages of the transition to digital imaging. Cameras using silver halide-based photographic film will soon be replaced by digital cameras, and a similar transition has begun from magnetic tape-based camcorders to digital video recorders. These transitions are, of course, coupled to the rapid advances in computer processing and storage technologies. With these transitions and computing advances will come an explosion not only in the number of stored digital images, but also in the number of applications that require them. Thus, as the prevalence of digital image data grows, so will the need for automating the process by which this data is enhanced, processed, or analyzed.
This group is studying methods for capturing, understanding, and predicting the appearance of the visual world. Success in this research domain necessitates a unified approach to open problems in two fields, namely computational vision and computer graphics. Our research effort is focused on a number of pertinent areas: sensing, modeling, estimation, generation, and evaluation. We are developing sensors that provide new types of visual information; complex models of materials, reflectances and textures; estimation algorithms that use our models to recover scene properties from minimal data; and advanced rendering techniques. We have placed particular emphasis on modeling the appearance of objects under varying lighting and on recovering the shape and reflectance properties of objects from both sparse and exhaustive photometric data. Applications for our research include face and object recognition, image-based rendering, video and image compression, visual tracking systems, object modeling systems, etc.