Object Representation and Feature Matching

Having established the utility of registration systems in developing complete 3-D models and their possible use in the pose estimation of matched objects, we now turn our attention to describing the representations and matching algorithms which have been employed by model-based recognition systems. As mentioned in Section 1.3.2, a model-based object recognition system is composed of two stages, the representational stage and the recognition stage (see Figure 1.2). The recognition stage is composed of two further sub-stages: matching and pose estimation.