Our research project on "Place Recognition Using Color Region Analysis" addresses the problem of determining the location of the represented scene in a known environment and also attempts to identify surfaces on an image based upon their color properties. More specifically, the goal of this identification process is to enable the department robot, Susan B., to answer the well-known computer vision question "Where am I?" The response to that question can provide indispensable information for root planning and location verification tasks.

The project investigates the efficiency of a characterizing feature that has not been examined extensively as a primary search feature in place recognition. It is suggested that color be used as an analysis attribute. The color measures of the identification processes are surface reflectance values. We were motivated by Edwin Land’s color constancy studies (introduced with the "Retinex Theorem") to give color a higher priority as a surface characterizing descriptor for three-dimensional objects and to use its power in the scene identification task. This research project was not only expected to provide an answer for the "Where am I?" problem, but also to confirm the effectiveness of Land’s reflectance algorithm in a more natural environment setting (e.g.: an office environment).

Throughout the localization procedure the input color images are compared to an environment model. This semantic network represents and describes the three-dimensional entities (locales) constituting the "real world" surroundings of an agent. The hierarchical structure among the locales reveals information about the relative location of these space volumes. The model also stores essential details about the geometrical properties and other characteristic features (e.g.: color) of surfaces, which are smaller building blocks of the environment.