This paper by Paul J. Besl and Ramesh C. Jain from the University of Michigan, Ann Arbor, focuses on the problem of three-dimensional (3-D) object recognition in computer vision. The authors define the 3-D object recognition problem and discuss the basic concepts and relevant literature. They also survey techniques for obtaining, processing, and characterizing range data, which are often used as sensor inputs instead of intensity images. The paper is structured into several sections: 1. **Problem Definition**: It provides a precise definition of 3-D object recognition, emphasizing the need for a system that can recognize objects from arbitrary viewpoints without prior knowledge of their structure. 2. **Mathematical Problem Formulation**: It formulates the problem mathematically using inverse-set mappings, detailing the relationship between depth-map functions and the possible objects that could have created them. 3. **Recognition System Components**: It outlines the components of a recognition system, including world modeling, image formation, description, understanding, and rendering processes. 4. **Characteristics of an Ideal System**: It lists the desirable characteristics of an ideal object recognition system, such as handling arbitrary viewing directions, complex objects, and noise. 5. **Literature Review**: It reviews existing literature on 3-D object representation schemes, surface representation schemes, rendering algorithms, image formation and processing, surface characterization, object reconstruction algorithms, and recognition systems using intensity and range images. The paper emphasizes the importance of surface characterization and matching algorithms in 3-D object recognition and suggests that a general-purpose vision system should be able to handle a wide range of objects and scenarios.This paper by Paul J. Besl and Ramesh C. Jain from the University of Michigan, Ann Arbor, focuses on the problem of three-dimensional (3-D) object recognition in computer vision. The authors define the 3-D object recognition problem and discuss the basic concepts and relevant literature. They also survey techniques for obtaining, processing, and characterizing range data, which are often used as sensor inputs instead of intensity images. The paper is structured into several sections: 1. **Problem Definition**: It provides a precise definition of 3-D object recognition, emphasizing the need for a system that can recognize objects from arbitrary viewpoints without prior knowledge of their structure. 2. **Mathematical Problem Formulation**: It formulates the problem mathematically using inverse-set mappings, detailing the relationship between depth-map functions and the possible objects that could have created them. 3. **Recognition System Components**: It outlines the components of a recognition system, including world modeling, image formation, description, understanding, and rendering processes. 4. **Characteristics of an Ideal System**: It lists the desirable characteristics of an ideal object recognition system, such as handling arbitrary viewing directions, complex objects, and noise. 5. **Literature Review**: It reviews existing literature on 3-D object representation schemes, surface representation schemes, rendering algorithms, image formation and processing, surface characterization, object reconstruction algorithms, and recognition systems using intensity and range images. The paper emphasizes the importance of surface characterization and matching algorithms in 3-D object recognition and suggests that a general-purpose vision system should be able to handle a wide range of objects and scenarios.