This paper presents a framework for representing and recognizing three-dimensional shapes, focusing on the design of a shape representation that is accessible, unique, and stable. The authors propose that a shape representation should be object-centered, use volumetric primitives, and have a modular organization. The representation should be based on the natural axes of the shape, such as those identified by a stick figure. The process of deriving a shape description involves identifying the natural axes of the shape in its image and transforming viewer-centered axis specifications into object-centered specifications. The paper discusses the criteria for evaluating the effectiveness of a shape representation, including accessibility, scope and uniqueness, and stability and sensitivity. It also explores the design aspects of a shape representation, such as the coordinate system, primitives, and organization. The authors argue that a shape representation should use an object-centered coordinate system, include volumetric primitives of varied sizes, and have a modular organization. This allows for a more efficient and accurate representation of shapes for recognition purposes. The paper also discusses the process of shape recognition, which involves a collection of stored shape descriptions and various indexes that allow a newly derived description to be associated with an appropriate stored description. The most important of these indexes allows shape recognition to proceed conservatively from the general to the specific based on the specificity of the information available from the image. New constraints supplied by a conservative recognition process can be used to extract more information from the image. A relaxation process for carrying out this constraint analysis is described. The paper also discusses the 3-D model representation, which is based on the principles of object-centered coordinate systems, volumetric primitives, and modular organization. The 3-D model representation is designed to capture the essential information about a shape in a way that is stable and sensitive to subtle differences. The paper also discusses the derivation of a 3-D model description from an image, the transformation of viewer-centered coordinates to object-centered coordinates, and the indexing of 3-D models for recognition purposes. The interaction between the derivation and recognition processes is also discussed, emphasizing the importance of using a conservative approach to ensure accurate and reliable shape recognition.This paper presents a framework for representing and recognizing three-dimensional shapes, focusing on the design of a shape representation that is accessible, unique, and stable. The authors propose that a shape representation should be object-centered, use volumetric primitives, and have a modular organization. The representation should be based on the natural axes of the shape, such as those identified by a stick figure. The process of deriving a shape description involves identifying the natural axes of the shape in its image and transforming viewer-centered axis specifications into object-centered specifications. The paper discusses the criteria for evaluating the effectiveness of a shape representation, including accessibility, scope and uniqueness, and stability and sensitivity. It also explores the design aspects of a shape representation, such as the coordinate system, primitives, and organization. The authors argue that a shape representation should use an object-centered coordinate system, include volumetric primitives of varied sizes, and have a modular organization. This allows for a more efficient and accurate representation of shapes for recognition purposes. The paper also discusses the process of shape recognition, which involves a collection of stored shape descriptions and various indexes that allow a newly derived description to be associated with an appropriate stored description. The most important of these indexes allows shape recognition to proceed conservatively from the general to the specific based on the specificity of the information available from the image. New constraints supplied by a conservative recognition process can be used to extract more information from the image. A relaxation process for carrying out this constraint analysis is described. The paper also discusses the 3-D model representation, which is based on the principles of object-centered coordinate systems, volumetric primitives, and modular organization. The 3-D model representation is designed to capture the essential information about a shape in a way that is stable and sensitive to subtle differences. The paper also discusses the derivation of a 3-D model description from an image, the transformation of viewer-centered coordinates to object-centered coordinates, and the indexing of 3-D models for recognition purposes. The interaction between the derivation and recognition processes is also discussed, emphasizing the importance of using a conservative approach to ensure accurate and reliable shape recognition.