Color indexing is a method for identifying and locating objects in real-time vision systems. This paper presents two algorithms: Histogram Intersection and Histogram Backprojection. Histogram Intersection is used for identifying objects by matching model and image histograms, while Histogram Backprojection is used for locating objects in crowded scenes. The paper argues that color is a robust and efficient cue for object identification and location, especially in dynamic environments. Color is local and largely independent of view and resolution, making it suitable for both tasks. The paper also discusses the role of color in vision, noting that while geometric cues are often considered more reliable for object identity, color may be more efficient for routine behavior. Color is used in natural and artificial environments to identify objects, and recent progress in color constancy algorithms suggests that color can be used as an invariant for object surfaces under varying light conditions. The paper also discusses the distinction between "what" (identification) and "where" (location) in vision, suggesting that the visual system processes these tasks separately. The paper concludes that color indexing can be used to build efficient, real-time vision systems for robots.Color indexing is a method for identifying and locating objects in real-time vision systems. This paper presents two algorithms: Histogram Intersection and Histogram Backprojection. Histogram Intersection is used for identifying objects by matching model and image histograms, while Histogram Backprojection is used for locating objects in crowded scenes. The paper argues that color is a robust and efficient cue for object identification and location, especially in dynamic environments. Color is local and largely independent of view and resolution, making it suitable for both tasks. The paper also discusses the role of color in vision, noting that while geometric cues are often considered more reliable for object identity, color may be more efficient for routine behavior. Color is used in natural and artificial environments to identify objects, and recent progress in color constancy algorithms suggests that color can be used as an invariant for object surfaces under varying light conditions. The paper also discusses the distinction between "what" (identification) and "where" (location) in vision, suggesting that the visual system processes these tasks separately. The paper concludes that color indexing can be used to build efficient, real-time vision systems for robots.