Welcome to the homepage of the MIT City Scanning Project Dataset. The purpose of this site is to make publicly available what we believe is the largest collection of calibrated ground imagery in existence. This dataset consists of a collection of roughly ten thousand geo-referenced, high dynamic range color images collected from a several hundred meter square outdoor region of the MIT campus. These images were collected using a digital camera mounted on a fully automated electromechanical pan-tilt head rotated about a fixed optical center into twenty orientations tiling a portion of the sphere. Each of these images is annotated (by on-board navigation instrumentation) with the time, date, and an estimate of the camera's absolute position and orientation in an Earth-centered, Earth-fixed coordinate system. After the images were collected they were processed using a suite of algorithms to recover intrinsic and extrinsic parameters more accurately. We are now validating the calibration data and posting it to this repository, along with the image and sub-pixel edge and point (edge intersection) features for each image.

      This dataset was gathered as part of the City Scanning Project lead by Seth Teller of the MIT Computer Graphics Group. This website is being developed by Zachary Bodnar as part of his M.Eng thesis.

      For more detailed information concerning the nature of this data please refer to the following document: Calibrated, Registered Images of an Extended Urban Area Teller, Antone, Bodnar, Bosse, Coorg, Jethwa, Master (also available in postscript).

      Note: This website makes heavy use of Java in a number of places. To get the best performance, you may want to consider trying Sun's latest Java run time environment: Java Plug-in 1.3

Browse the repository:	browse
Register to access the repository online:
Return users may login here:

      This repository is still under development. We would appreciate any feedback you might have.

A few sample nodes from the database

      Support for this research was provided in part by a Sloan Foundation Research Fellowship, in part by the Office of Naval Research under MURI Award SA 1524-2582386, in part by Intel Corporation, in part by Draper Laboratories under the IRAD program, in part by Lincoln Laboratories under the ACC program, and in part by the NTT Corporation under Award MIT9904-20.