This paper surveys content-based 3D shape retrieval methods, discussing their applicability to both surface and volume models. It evaluates methods based on requirements such as shape representation, dissimilarity measures, efficiency, discrimination ability, partial matching, robustness, and pose normalization. The paper discusses various aspects of 3D shape retrieval, including shape representations, similarity measurement, efficiency, discriminative power, partial matching, robustness, and pose normalization. It categorizes shape matching methods into feature-based, graph-based, and other methods. Feature-based methods use geometric and topological properties of 3D shapes, while graph-based methods use graph structures to represent shapes. Other methods include view-based similarity, volumetric error-based similarity, weighted point set-based similarity, and deformation-based similarity. The paper concludes that feature-based methods are generally robust and efficient, while graph-based methods have limited discriminative power. It identifies research issues such as benchmark comparisons and the need for more efficient methods.This paper surveys content-based 3D shape retrieval methods, discussing their applicability to both surface and volume models. It evaluates methods based on requirements such as shape representation, dissimilarity measures, efficiency, discrimination ability, partial matching, robustness, and pose normalization. The paper discusses various aspects of 3D shape retrieval, including shape representations, similarity measurement, efficiency, discriminative power, partial matching, robustness, and pose normalization. It categorizes shape matching methods into feature-based, graph-based, and other methods. Feature-based methods use geometric and topological properties of 3D shapes, while graph-based methods use graph structures to represent shapes. Other methods include view-based similarity, volumetric error-based similarity, weighted point set-based similarity, and deformation-based similarity. The paper concludes that feature-based methods are generally robust and efficient, while graph-based methods have limited discriminative power. It identifies research issues such as benchmark comparisons and the need for more efficient methods.