In this paper we described a method to correct errors on a three orthographic views including contradictions in a system that reconstructs 3-D models of polyhedrons. When a contradiction with respect to attribute of segments (solid line, dotted line and chain line) is detected in the process of 3-D model reconstruction, search space is expanded by reducing constraints that are used in searching process in order to continue searching. 3-D model is reconstructed with attribute of segments corrected. The function of this method is confirmed by applying it to three orthographic views including curved surfaces.

View full abstract

We have been investigating the automatic reconstruction of solids from a set of three views, In the case of one or two views employed in the conventional drawing manner, the deal of information from projections decreases and the number of the candidate solids reconstructed geometrically increases, When we formulate the tacit understanding for which the object uniquely corresponds to the projections in the drawing manner, we can develop the reconstruction method from two views. In this paper, we propose the algorithm to reconstruct automatically from two views and show the application to the machine parts.

View full abstract

Existing CAD systems do not provide advanced functions for automatic checking of design and drafting errors in mechanical drawings. If the knowledge of checking of mechanical drawings can be implemented into computers, CAD systems could automatically check for design and drafting errors. This paper describes a method for systematic check-ing of dimensional errors, such as deficiency and/or redundancy of dimensions, input-errors in dimension entities (texts, figures, and symbols), etc. The logic for finding dimensional errors is written by using a procedural language C^<++>. A geometric model and a topological-graph model are used in this method. Checking for deficiency and redundancy of dimensions is based upon Graph Theory.

View full abstract

This report presents a feature-based approach for the computer processing of machine drawings, which can be used to the machining process planning. In the stage of solid object generation, all loops in a three-view drawing are divided into primitive loops and every primitive solid is generated from three primitive loops, each of which is selected separately from every view drawing. The collection of these primitive solids form a solid object corresponding to the input drawing. Considering the difference between two collections of primitive solids for the original drawing and for the base stock drawing, a machining process can be planned.

View full abstract

This paper describes a method and test collections for constructing a solid model from CAD data of drawings. At present, design and manufacturing are guided and carried out mainly by Fleshing out three dimensional drawings, and most of drawings are input by CAD systems. models from drawings enables more effective use of valuable drawings as well as enables us However, drawings would not to receive merits given by three demensional CAD systems. actually give full information for automatic building of three dimensional models and the method should be developed for further application. Authors have been doing research for constructing three dimensional models from drawings, and <<Unkei>> system has been developed. Some examples are illustrated and key points for applying this technology for industry are explained.

View full abstract

This paper presents a method to reconstruct a CSG solid model, which is composed of a binary tree of direct sum and difference operations, from a set of orthographic three views of a mechanical drawing. The main stages of the method are. (1) to extract all (under certain constraints) combinations of the loops of 2D primitives for each view, (2) to generate all possible combinations of 3D primitives and (3) to finally generate a CSG tree of direct sum and difference operations. The algorithm is described clearly in detail and some examples are finally shown.

View full abstract

This paper discusses an attempt to develop a system that automatically reconstructs a three dimensional (3D) solid model from three orthographic views. The CSG method was adopted for 3D reconstruction where the basic idea is the following: a 3D solid model is constructed by geometric combinations of simple 3D solids, called primitives. Appropriate primitives are chosen by comparing their projection with the original three views that are augmented and segmented by a rule proposed here. The chosen primitives are combined according to a structure tree. It represents primitives' size and inclusion relationships and is constructed from the original views. The proposed system was successfully tested using several simple but practical examples. With some examples, however, some difficulties were observed to which we propose alternate approaches.

View full abstract

Much research has been conducted on generating assembly sequences from assembly drawings on CAD systems. However, methods proposed previously have difficulties in generating assembly sequences when assemblies include complex cases, and have limitations of being applied to general cases. This paper describes an algorithm that generates assembly sequences of assemblies that include nonlinear, nonmonotonic, noncoherent assemblies. A point of the algorithm is to make conjunctions of directions of parts along which the parts could move with respect to other parts, and to establish a set of equations of the geometric relationships in order to calculate the conjunctions.

View full abstract

A surface model representation of a solid can be constructed in straightforward fashion from its orthographic views. Such a representation, however, may include ghost vertices, ghost edges and ghost faces. In the basis on a set of valid topological / geometrical rules dominating the surface model representation of solids, such ghosts can be removed by the formal constraint satisfaction problem by "Pseudo Boolean Algebraic Nonlinear Programming". This method solves a system of simultaneous equations based on the above rules for the surface model representation with ghosts, and it produces a true surface model representation without ghosts. This paper describes the processing flow, formalization and applicability of our methods with use of some examples. All polyhedra satisfying the orthographic views can be constructe by solving the simultaneous equations on the constraint part of nonlinear integer programming. A unique polyhedron can be constructed from the ambiguous orthographic views by adding objective functions to the nonlinear integer programming. This objective functions are based on heuristics.

View full abstract