A review on strategies for recognizing natural objects in colour images of outdoor scenes

This paper surveys some significant vision systems dealing with the recognition of natural objects in outdoor environments. The main goal of the paper is to discuss the way in which the segmentation and recognition processes are performed: the classical bottom-up, top-down and hybrid approaches are discussed by reviewing the strategies of some key outdoor scene understanding systems. Advantages and drawbacks of the three strategies are presented. Considering that outdoor scenes are especially complex to treat in terms of lighting conditions, emphasis is placed on the way systems use colour for segmentation and characterization proposals. After this study of state-of-the-art strategies, the lack of a consolidated colour space is noted, as well as the suitability of the hybrid approach for handling particular problems of outdoor scene understanding. (C) 2000 Elsevier Science B.V. All rights reserved.