Mathematical formulations and procedures for computing structural constraints between 3D geometrical entities are presented. Geometrical entities concerned in this paper include point, line segment, and polygon, which are easy and reliable to be detected from sensory data and are preferable to be selected as features for object recognition. Structural constraints describing relationships among geometrical entities provide shape information of objects by which objects can be discriminated. Two kinds of structural constraints are considered: distance constraints and angle constraints. Each kind of constraints is further classified, in terms of the combination of entity types, into finer groups. There are six groups of distance constraint and three groups of angle constraint. Their mathematical formulations are collected and experienced. Areas of application and extension are outlined as well.