An adaptive image preprocessing model for classifying drawings of building metal structures

The constant growth of the engineering services market and the significant volume of archival paper documentation make the task of digitizing technical drawings and their specifications urgent. An analysis of the current state of research demonstrates the active use of recognition and generative models to extract key drawing characteristics, which facilitates their integration into computer modeling systems. Image preprocessing is used to improve the efficiency of neural network models; however, due to the specific structure and potential sources of drawings, this approach carries the risk of information loss or data corruption. The authors propose an adaptive preprocessing model designed for classification systems for drawings of building metal structures. This model includes scaling, noise filtering, binarization, and color restoration of informative pixels, where each operation is applied depending on the values of criteria that determine whether it is required. The research identifies algorithms and methods suitable for implementing the above-described operations: the Lanczos interpolation method, the switching weighted median filter, and the Wolf binarization method. The authors propose a set of criteria for evaluating the effectiveness of preprocessing in classifying drawings of building metal structures. A series of experiments shows that the developed adaptive preprocessing model yields a greater increase in classification accuracy compared to standard preprocessing based on the same algorithms or no preprocessing at all. Conclusions are drawn regarding the potential for further implementation of the model in systems for classification and recognition of parts in drawings of building metal structures.

1. Lu T., Yang H., Yang R., et al. Automatic analysis and integration of architectural drawings. International Journal of Document Analysis and Recognition. 2007;9(1):31–47. https://doi.org/10.1007/s10032-006-0029-6

2. Zhang L.Y., Zhu J.Y., Zhu Q.Y. A New System for Automatic Understanding Engineering Drawings. CIRP Annals. 1992;41(1):477–479. https://doi.org/10.1016/S0007-8506(07)61248-7

3. Fan N. Application of CAD Combined with Computer Image Processing Technology in Mechanical Drawing. In: Application of Intelligent Systems in Multi-modal Information Analytics: Proceedings of the 2020 International Conference on Multi-model Information Analytics (MMIA2020): Volume 2, 05–06 March 2020, Changzhou, China. Cham: Springer; 2021. P. 319–324. https://doi.org/10.1007/978-3-030-51556-0_46

4. Bai Y.B., Xu X.W. Object Boundary Encoding – A New Vectorisation Algorithm for Engineering Drawings. Computers in Industry. 2001;46(1):65–74. https://doi.org/10.1016/S0166-3615(01)00115-4

5. Lin H.-Y., Lin Ch.-Y., Lin Ch.-J., et al. A Study of Digital Image Enlargement and Enhancement. Mathematical Problems in Engineering. 2014;2014:825169. https://doi.org/10.1155/2014/825169

6. Keys R. Cubic Convolution Interpolation for Digital Image Processing. IEEE Transactions on Acoustics, Speech, and Signal Processing. 1981;29(6):1153–1160. https://doi.org/10.1109/TASSP.1981.1163711

7. Zhu Y., Dai Y., Han K., et al. An Efficient Bicubic Interpolation Implementation for Real-Time Image Processing Using Hybrid Computing. Journal of Real-Time Image Processing. 2022;19(6):1211–1223. https://doi.org/10.1007/s11554-022-01254-8

8. Chunshi L., Yu L., Yimin Zh. Study of B-Spline Interpolation, Correction and Inverse Algorithm. In: Advances in Automation and Robotics: Selected papers from the 2011 International Conference on Automation and Robotics (ICAR 2011): Volume 2, 01–02 December 2011, Dubai, UAE. Berlin, Heidelberg: Springer; 2012. P. 215–221. https://doi.org/10.1007/978-3-642-25646-2_30

9. Simon H.D. The Lanczos Algorithm with Partial Reorthogonalization. Mathematics of Computation. 1984;42(165):115–142. https://doi.org/10.2307/2007563

10. Boyat A.K., Joshi B.K. A Review Paper: Noise Models in Digital Image Processing. Signal & Image Processing: An International Journal. 2015;6(2):63–75. https://doi.org/10.5121/sipij.2015.6206

11. Fatemi M., Amindavar H., Ritcey J.A. Noise reduction via harmonic estimation in Gaussian and non-Gaussian environments. Signal Processing. 2010;90(5):1554–1561. https://doi.org/10.1016/j.sigpro.2009.11.002

12. Haralick R.M., Sternberg S.R., Zhuang X. Image Analysis Using Mathematical Morphology. IEEE Transactions on Pattern Analysis and Machine Intelligence. 1987;PAMI-9(4):532–550. https://doi.org/10.1109/TPAMI.1987.4767941

13. Sun M. Comparison of Processing Results of Median Filter and Mean Filter on Gaussian Noise. Applied and Computational Engineering. 2023;5(1):779–785. https://doi.org/10.54254/2755-2721/5/20230702

14. Vo G.D., Park Ch. Robust Regression for Image Binarization under Heavy Noise and Nonuniform Background. Pattern Recognition. 2018;81(4):224–239. https://doi.org/10.1016/j.patcog.2018.04.005

15. Otsu N. A Threshold Selection Method from Gray-Level Histograms. IEEE Transactions on Systems, Man, and Cybernetics. 1979;9(1):62–66. https://doi.org/10.1109/TSMC.1979.4310076

16. Kapur J.N., Sahoo P.K., Wong A.K.C. A New Method for Gray-Level Picture Thresholding Using the Entropy of the Histogram. Computer Vision, Graphics, and Image Processing. 1985;29(3):273–285. https://doi.org/10.1016/0734-189X(85)90125-2

17. Sauvola J.J., Pietikäinen M. Adaptive Document Image Binarization. Pattern Recognition. 2000;33(2):225–236. https://doi.org/10.1016/S0031-3203(99)00055-2

18. Wolf C., Jolion J.-M. Extraction and recognition of artificial text in multimedia documents. Formal Pattern Analysis & Applications. 2004;6(4):309–326. https://doi.org/10.1007/s10044-003-0197-7