Development of a face verification system for a home service robot

The robot providing services based on interaction with a human needs a fast verification process that knows whether the user belongs to a family or a guest group for providing the differentiated services to each group member. In this paper, we developed a verification system for one-to-group matching (e.g., user-to-family or user-to-guest) and tested on the database acquired using the robot in an uncontrolled environment. The proposed system consists of three parts: face detection using revised modified census transform and adaboost, feature extraction using multiple principle component analysis, and verification using the aggregating Gaussian mixture model-universal background model. Experimental results show that the proposed system works well on the deteriorated database with a high speed. (C) Koninklijke Brill NV, Leiden and The Robotics Society of Japan, 2008.