| Christopher Brown (Professor ) |
| UR > CS > Research > Vision > People > Christopher Brown |
![]() | Contact:Christopher BrownChris Brown Department of Computer Science Computer Studies Building -- Room 734 160 Trustee Road University of Rochester Rochester, NY 14627-0226 vox: (585) 275-7852 fax: (585) 273-4556 email: brown@cs.rochester.edu home: http://www.cs.rochester.edu/u/brown/ vision: http://www.rochester.edu/research/vision/people/Christopher_M_Brown |
About:Chris Brown (BA Oberlin 1967, PhD U. Chicago 1972) is Professor of Computer Science at the University of Rochester. He has been at Rochester since finishing a postdoctoral fellowship at the School of Artificial Intelligence at the University of Edinburgh in 1974 (the leading European research center of that time). With his Rochester colleague Dana Ballard, he is coauthor of the leading textbook in the field, COMPUTER VISION. He spent two years heading the software development of the PADL-2 solid modeling package from the Production Automation Project at Rochester. He was chairman of his department from 1981-1984. His current research interests are augmented reality, computer vision and robotics, especially integrated parallel systems performing animate vision (the interaction of visual capabilities and motor behavior). The vision work spans a range of topics from traditional symbolic artificial intelligence to low-level aspects of control and includes assembling hardware and software support systems. He is the faculty advisor of the Computer Interest Floor and the titular supervisor of the competition-winning UR undergraduate robotics team. In his copious spare time he cooks, brews beer, hunts, pursues classic guitar studies, and revels in the "joys of home ownership". | |
Research Interests:Some of my current activities center around building behaving, visually guided systems. Traditionally, much of the effort is interdisciplinary, involving symbolic AI and parallel computing, and is carried out jointly with my colleagues Nelson and Ballard, and in collaboration with the planning research group (Allen, Schubert, and Kyburg) and the Systems group (Scott, Murphy, Ding, Shen). The vision work spans a range of topics from traditional symbolic artificial intelligence to low-level aspects of control and includes assembling hardware and software support systems. Our work on gaze control has explored the cooperation of different control mechanisms (for example, those that track objects and those that verge the eyes), both to improve tracking performance and to render early visual operations more robust in dynamic scenes. Learning is a basic capability that relieves the AI programmer from the impossible burden of anticipating all possible situations. Learning is a very active topic at Rochester. Early work on learning control structures, skilled action sequences, and three-dimensional object representations is described in the November 1991 issue of the International Journal of Computer Vision, which is devoted to vision research at Rochester. More recent work involves learning control strategies for skilled sensori-motor tasks. The high-level, cognitive control of scarce visual resources (such as deciding where to point a narrow-angle camera in a wide-angle scene) is a vital topic for automatic systems, which (like humans) have no hope of visually analyzing everything continuously. Early on, we used Bayesian probability theory and decision analysis to develop a theory of selective visual attention and task-oriented vision (the "where to look next" problem) , and to direct our binocular robot head. In 2000, Dr. Joachim Denzler of Erlangen spent a year here and we did work in "how to look next", or optimal camera parameter selection, which included both internal and external camera parameter settings. Currently, probabilistic reasoning is used for interpreting consumer photos, in joint work with Eastman Kodak Company carried out by Matt Boutell. Augmented reality (mixing real-time graphics and artificial haptics with live video) raises several technical problems, and over the past decade we've been working on aspects of this area. Most recently we have working to render more faithfully the "physical" interactions of real and graphically simulated objects occupying the same (augmented) scene. This work is being done by Bo Hu. The UR undergraduate robotics team is one of several undegraduate research projects in CS. The roboteers won the host robot competition in Edmonton, AL in August of 2002 with no help from me, but I'm the official sponsor and associated academic sponsor. Collaboration with forward-looking industry continues with AppleAid Inc., a local research-oriented firm doing robotics (we're working on navigation and obstacle avoidance), and most recently with the thesis of Chris Eveland (Dec. 2002), who has been working at Equinox Corp. on special sensors (polarization and long- and medium-wave IR). | |
Projects:Improving Augmented Reality Rendering: We improve the human-computer interaction experience in AR systems by simulating global lighting effects between virtual and real objects.Perceptual Basis of Spatial Relations: We seek to visually recognize the spatial relations between objects in a 3-D realistic environment, using an interaction-based approach that identifies a small high-leverage set of visual features. Quasi-static Object Discovery (QsOD): We discover quasi-static objects, objects that are stationary during some interval of observation, across image sequences acquired by any number of completely uncalibrated cameras using only temporal (no spatial) information. Semantic Scene Classification: We improve semantic scene classification by using semantic object/ material detectors and spatial models within a probabilistic framework to infer the scene type. Mabel (the Mobile Table): Mabel(the Mobile Table) is a robotic system that can perform waypoint and vision guided navigation, speech generation, speech recognition, person finding, face finding, and face following. | |
Publications:Tutorial on Filtering, Restoration, and State Estimation (Christopher M. Brown)Vision, Learning, and Development (Christopher Brown) Design and Evaluation of a System for Vision-Based Vehicle Convoying (Rodrigo L. Carceroni, Craig Harman, Christopher K. Eveland and Christopher M. Brown) Information Theoretic Sensor Data Selection for Active Object Recognition and State Estimation (J. Denzler and C.M. Brown) Interactive Indoor Scene Reconstruction from Image Mosaics using Cuboid Structure (Bo Hu and Christopher M. Brown) {Optimal Camera Parameter Selection for State Estimation with Applications in Object Recognition} (J. Denzler and Christopher M. Brown and H. Niemann) Acquiring an Environment Map through Image Mosaicking (Bo Hu and Christopher M. Brown and Andrew Choi) Dynamic Bayes Net Approach to Multimodal Sensor Fusion (Amit Singhal and Chris Brown) Numerical Methods for Model-Based Pose Recovery (Rodrigo L. Carceroni and Christopher M. Brown) Mobile Robot (Jessica D. Bayliss and Christopher M. Brown and Rodrigo L. Carceroni and Christopher K. Eveland and Craig Harman and Amit Singha and Mike Van Wie) Goal-Orianted Dynamic Vision (Peter A. von Kaenel and Christopher M. Brown and Raymond D. Rimey) A Fully Projective Formulation for Lowe's Tracking Algorithm (Araujo Helder and Rodrigo L. Carceroni and Christopher M. Brown) Decoupling Orientation Recovery from Position Recovery with 3D-2D Point Correspondences (Rodrigo L. Carceroni and Christopher M. Brown) Robot Skill Learning and the Effect of Basis Function Choice (J.G. Schneider and Christopher M. Brown) Optimal Selection of Camera Parameters for State Estimation of Static Systems: An Information Theoretic Approach (Joachim Denzler and Christopher M. Brown) | |