A new method for identifying and validating features from 2D sectional views

Abstract

Feature identification is one of the key steps for 3D solids reconstruction from 2D vector engineering drawings using the volume-based method. In this paper, we propose a novel method to identify and validate features from sectional views. First, features are classified as explicit features (EPFs) and implicit features (IPFs), which are then identified in an order of priority using heuristic hints. We show that the problem of constructing EPFs can be formulated as a 0-1 integer linear program (ILP), and the IPFs are generated based on the understanding of semantic information of omitted projections in sectional views. Then, the Loop-Relation Graph (LRG) is introduced as a multi-connected-subgraph representation for describing the relations between loops and features. According to the LRG, a reasoning technique based on confidence is implemented to interactively validate features. This method can recover features without complete projections, and the level of understanding sectional views is improved. Full sections, partial sections, offset sections as well as revolved sections can be handled by our method. Several examples are provided to demonstrate the practicability of our approach.

Introduction

Engineering drawings have been used as a standard and unified language for describing mechanical designs since the 19th century [1]. Now they still play an important role in engineering practices since many product designs are definitively expressed by means of 2D engineering drawings. Moreover, the manufacturing, fabrication and assembly of products can be guided by 2D engineering drawings. However, 3D solid models have become more useful mechanical CAD tools since they are easier to visualize and modify in downstream processes [2]. Considering that the most existing product design documentations are represented in the form of 2D engineering drawings, it is necessary to convert 2D engineering drawings to 3D solid models for downstream computer-aided manufacturing techniques, such as finite-element analysis, process planning, numerically controlled machining and emulation display [3].

There are two main formats for the existing engineering drawings: digitally scanned format and vector format. Some early engineering drawings are drawn on papers. These drawings have been kept and handed down in the format of digitally scanned images. Since 1990s, vector engineering drawings have become popular because many CAD softwares (e.g. AutoCAD) are widely used by engineers to carry out design work. Digitally scanned engineering drawings can be effectively transformed into vector drawings by vectorization, which has been the subject of much research. Therefore, in this paper we only focus on the reconstruction of vector engineering drawings.

Recent thirty years have witnessed significant progress in reconstructing 3D solid models from 2D vector engineering drawings. The existing reconstruction methods can be basically classified into two categories [4], [5]: wireframe-oriented approach and volume-based approach. The wireframe-oriented approach involves four main steps. (1) Transforming 2D junctions to 3D vertices; (2) generating 3D edges from 3D vertices; (3) constructing 3D faces from 3D edges and (4) forming 3D objects from 3D faces. This approach was first proposed by Idesawa [6] and then was formalized by Markowsky and Wesley [7], [8]. Later, many researchers have extended the domain of objects of this approach from polyhedrons [9], [10], [11] to quadric surfaces [3], [12]. The volume-based approach creates elementary solids by recognizing patterns in 2D projections [13], [14], or takes 2D loops as bases and uses extrusion [15], [16] or rotation [17], [18], [19] to construct 3D elementary objects, and then assembles them to obtain the 3D models. Recently, some unconventional algorithms have been proposed. Ibrahim [20] considered the original object as a prismatic raw material and the final object was obtained by removing features that were identified by his 3-Space method. Zhang [21] presented a new reconstruction framework, which accounts for both the engineering drawings themselves and the process planning information. Series of 3D models can be constructed by simulating the course of manufacturing mechanical parts. However, only rotational parts can be handled by this algorithm.

Current studies mainly address three orthographic views. In the conventional orthographic views, dashed lines are used to represent the hidden part from the view direction, making projections of an object with intricate interior details too complex to understand. Compared with the orthographic views, sectional views are easier to draw and read [2], [22]. Sectional views can describe complicated objects more clearly by using one or more planes to cut the objects and remove the parts obstructing the observer. Therefore, sectional views are more commonly used than three orthographic views. Many different types of sectional views are described and listed in [2], [22].

In engineering practices, most actual drawings contain sectional views [23]. However, it is difficult for computers to understand sectional views since many edges are omitted. In this paper, our purpose is to solve this issue. Based on the understanding of semantic information in sectional views, we present a new feature identification and validation algorithm. First, a new method is proposed to directly recognize explicit features (EPFs) as well as implicit ones (IPFs) by a hint-based search strategy. Features whose parts of projections are omitted can be identified and constructed by this method. Then, the Loop-Relation Graph (LRG), which represents the relations between loops and features, is implemented to interactively validate features. By the LRG, the unordered features are classified into several groups and assigned with confidence values. The users can select final features according to the confidence values. Using our method, we extend the domain of features that can be identified automatically, and improve the level of understanding sectional views with incomplete projections.

The rest of the paper is organized as follows. The related work about reconstruction of sectional views is introduced in Section 2. In Section 3, the definitions and terminology used in our algorithm are given. Section 4 is our feature identification algorithm. Section 5 describes the feature validation algorithm. Experimental results are presented in Section 6. Finally, in Section 7 we come to the conclusions and describe our future work.