ANALYSIS OF THE MAIN METHODS OF DRAWINGS RECOGNITION AND THE POSSIBILITIES OF TRANSFORMATION 2D IN 3D
DOI:
https://doi.org/10.28925/2663-4023.2022.16.185193Keywords:
2D; 3D; recognition; artificial neural networks; drawing; two-dimensional models; three-dimensional models.Abstract
The article presents an analysis of the main methods for recognizing drawings and the possibilities of transforming two-dimensional models (2D) into three-dimensional models (3D). Despite the rapid development of IT, the question of accuracy and speed of transformation of two-dimensional models into three-dimensional ones remains open. As machine design technologies and corresponding automated decision-making systems (CAD) develop, the number of methods and models that can potentially be used in the task of drawing recognition and 2D to 3D transformation is rapidly increasing. Today, there are quite a large number of methods for recognizing drawings and converting them into a three-dimensional model, but each of them has a certain number of shortcomings. Therefore, there is a need to carry out a comprehensive analysis of these methods, which can potentially be applied in the context of solving problems of drawing recognition and 2D to 3D transformation.
It should be noted that there is a contradiction between the traditional procedure for preparing drawing documentation on paper media until the 80s and 90s of the 20th century and the new methods of 3D modelling that have been developed since the mid-90s. This gives designers truly unlimited opportunities to prepare design and technical documentation, without focusing on the problem of preparing design and drawing documentation and the features of entering input data. Application software significantly facilitates this process. Note that most 3D systems (for example, software products Autodesk TinkerCAD, DesignSpark Mechanical, FreeCAD, Autodesk AutoCAD, ZBrush, Blender, etc.) use approaches that allow synthesizing a frame or boundary representation of an object modelled in space. Professional systems (for example Autodesk 3ds Max, Autodesk Maya) use generalized models of spatial objects. This idea assumes that the designers have appropriate information arrays, which a priori should correspond to all projections of the object in the three main planes.
Downloads
References
Sutherland, I.E. (1963). SCETCHPAD: A Man-Machine Graphical Communication System. Proc. SJCC 23.
Thornton, R.W. (1978). Interactive Modelling in Three Dimensions through Two-Dimensional Windows. Third International Conference and Exhibition on Computer in Engineering and Building Design.
Idesawa, M. (1973). A system to generate a solid figure from three view. Bulletin of JSME, 16(92), 216–225. https://doi.org/10.1299/jsme1958.16.216.
Woo, T.C., Hammer, J.M. (1977) Recognition of Three Dimensional designs from orthographic projections. In Proc. 9th CIRP Conference: Cranfield Institute of Technology, (pp. 247 – 255)
Preiss, K. (1980). Constructing the 3D representation of a plane-faced object from a digitized engineering drawing. Computer-Aided Design, 12(2), 83. https://doi.org/10.1016/0010-4485(80)90481-9.
Preiss, K. (1984) Constructing the Solid representation from engineering projections. Computer & Graphics,8(4), 381 –389.
Haralick, R.M., Queeney, D. (1982) Understanding engineering drawings. Comp. Graphics and Image Processing, 20(3), 244 – 258.
Markowsky, G., Wesley, M.A. (1981). Fleshing out projections. IBM J. Res& Develop, 25(6), 934 – 954.
G Markowsky, G., Wesley, M.A. (1980). Fleshing out wire frames. IBM J. Res.& Develop, 24(5), 582 – 587.
Markowsky, G., Wesley, M.A. (1986). Generation of solid models from two-dimensional and three-dimensional data. In Pickett, MS and Boyse, J M (eds). Solid modelling by computer: from theory to application: Plenum, 23 – 51.
Sakurai, H., Gossard, D.C. (1983) Solid Model Input Through Orthographic Views. Computer Graphics, 17(3), 243 – 252.
Wang, Z., Tsumura, K., Saito, Y. (1994). Recognition of hand-written mechanical drawing by multi-level neural network (MLNN). In Advancement of intelligent production (pp. 1–6). Elsevier. https://doi.org/10.1016/b978-0-444-81901-7.50015-3
Elyan, E., Jamieson, L., Ali-Gombe, A. (2020). Deep learning for symbols detection and classification in engineering drawings. Neural Networks, 129, 91-102. https://doi.org/10.1016/j.neunet.2020.05.025.
Zehtaban, L., Elazhary, O., Roller, D. (2016). A framework for similarity recognition of CAD models. Journal of Computational Design and Engineering, 3, 274-285. https://doi.org/10.1016/j.jcde.2016.04.002.
Wang, Z., Latif, M. (2007). Reconstruction of 3D Solid Models Using Fuzzy Logic Recognition. Proceedings of the World Congress on Engineering, 1, 37-42.
Kargas, A., Cooley, P., Richards, T. H. E. (1988). Interpretation of engineering drawings as solid models. Computer-Aided Engineering Journal, 5(2), 67. https://doi.org/10.1049/cae.1988.0016.
(1988). EUCLID. Technical Report. MATRA Datavision.
Omura, G. (2012). Mastering AutoCAD 2013 and AutoCAD LT 2013. John Wiley & Sons.
Avetisyan, A., Dahnert, M., Dai, A., Savva, M., Chang, A. X., Nießner, M. (2019). Scan2cad: Learning cad model alignment in rgb-d scans. In Proceedings of the IEEE/CVF Conference on computer vision and pattern recognition (pp. 2614-2623).
https://doi.org/10.48550/arXiv.1811.11187
Yuan, S., Chan, H. C. S., Filipek, S., Vogel, H. (2016). PyMOL and inkscape bridge the data and the data visualization. Structure, 24(12), 2041–2042. https://doi.org/10.1016/j.str.2016.11.012
