MODIFICATION OF THE HOUGH ALGORITHM FOR RECOGNITION AND CLASSIFICATION OF LINE IMAGES

UDC 004.93

 

Talapina Darya Mikhailovna – Master's degree student, the Department of Informatics and Web-design. Belarusian State Technological University (13a, Sverdlova str., 220006, Minsk, Republic of Belarus). E-mail: taladarmi@mail.ru

Novoselskaya Olga Aleksandrovna – PhD (engineering), Assistant Professor, the Department of Informatics and Web Design. Belarusian State Technological University (220006, Minsk, Sverdlova str., 13a, Republic of Belarus). E-mail: nochka@tut.by

DOI: https://doi.org/ 10.52065/2520-6141-2024-284-12.

Key words: image recognition; line art; graphic primitives; Hough algorithm; image pre-processing; image post-processing, Euclidean distance.

For citation: Talapina D. M., Novoselskaya O. A. Modification of the Hough algorithm for recognition and classification of line images. Proceedings of BSTU, issue 3, Physics and Mathematics. Informatics, 2024, no. 2 (284), pp. 89–95 (In Russian). DOI: 10.52065/2520-6141-2024-284-12.

Abstract

The article considers the question of recognition the complex for analysis non-uniform lines. To select a traditional algorithm for feature extraction, an analysis of existing algorithms is performed. Since the line image will consist of geometric figures in the form of non-uniform strokes of different thickness, the Hough algorithm, which is a traditional algorithm of computer vision, is taken as a basis. It is known that the Hough algorithm has a number of disadvantages. In particular, there are problems of recognition of dashed lines, as well as lines of limited length. The article shows the process of developing an algorithmfor recognition of line images aimed at analysis, interpretation and classification of line images and graphic primitives presented in the form of non-uniform lines. The features of representation of line images are considered, the requirements for the recognition algorithm are summarized. The developed algorithm is based on a modified Hough algorithm adapted for recognition of dashed, solid and dash-dotted lines. The article describes the main stages of the algorithm, including image preprocessing, definition and classification of graphic primitives, and postprocessing methods. Based on the analysis, it is advisable to use a combination of convolution filters (Gaussian filter and Cani filter) as preprocessing algorithms. It was decided to use a morphological dilation operation to expand light spaces in the image. The features of the modified algorithm proposed by the authors are shown. Calculation of line parameters using Euclidean distance is carried out. Effectiveness of the developed algorithm is assessed. The key feature of the proposed method is the ability to output the parameters of dashed images in a separate file.

Download

References

  1. Nazdryukhin A. S., Khramtsov I. N., Tushev A. N. Processing images of sales receipts for highlighting and recognizing text information Vestnik Dagestanskogo gosudarstvennogo tekhnicheskogo universiteta. Tekhnicheskiye nauki [Bulletin of the Dagestan State Technical University. Technical science], 2019, vol. 46, no. 4, pp. 113–122 (In Russian).
  2. Tsepkovskaya T. A. Questions of search and classification of graphic primitives inside analyzed images. Perspektivnoye razvitiye nauki, tekhniki i tekhnologiy [Perspective development of science, technology and technology], 2020, pp. 268–271 (In Russian).
  3. Bolgov A. N., Favorskaya M. N. Application of the Hough transform and its modifications for finding graphic primitives. Aktual'nyye problemy aviatsii i kosmonavtiki [Current problems of aviation and astronautics], 2011, vol. 1, no. 7, p. 352 (In Russian).
  4. Favorskaya M. N. Hough transform for recognition problems. DSPA: Voprosy primeneniya tsifrovoy obrabotki signalov [DSPA: Issues of application of digital signal processing], 2016, vol. 6, no. 4, pp. 826–830 (In Russian).
  5. Kudrina M. A. Using the Hough transform to detect straight lines and circles in an image. Izvestiya Samarskogo nauchnogo tsentra RAN [News of the Samara Scientific Center of the Russian Academy of Sciences], 2014, no. 4–2, pp. 476–478 (In Russian).
  6. Ovchinnikov A., Fan Chan Dang Ho. Processing of hemispherical images: determining the center of the image and highlighting vertical lines. Komponenty i tekhnologii [Components and Technologies], 2011, no. 12, pp. 133–136 (In Russian).
  7. Panfilova E., Kunina I. Using the windowed Hough transform to search for extended boundaries in an image. Sensornyye sistemy [Sensor systems], 2020, no. 34, pp. 340–353. DOI: 10.31857/S0235009220030075 (In Russian).
  8. Matas J., Galambos C., Kittler J. Progressive Probabilistic Hough Transform. Proc. IEEE Computer Society Conference on Computer Vision and Pattern Recognition. 1999, vol. 1.
  9. Mammeri A., Bouckerche A., Lu G. Lane detection and tracking system based on the MSER algorithm, Hough trans-form and Kalman filter. Proceedings of the 17th ACM international conference on Modeling, analysis and simulation of wireless and mobile systems. 2014, pp. 259–266.
  10. Solov’yov F. S., Tarasov I. E., Petrov A. B. Pattern recognition and detection of object contours in an image. Innovatsii v nauke [Innovations in science], 2019, no. 4 (92), pp. 4–9 (In Russian).
  11. Astratov O. S., Filatov V. N., Kuzmin S. A. Modified Hough transform in a computer vision system. Uspekhi sovremennoy radioelektroniki [Advances in modern radio electronics], 2013, no. 2, pp. 77–83 (In Russian).
  12. Denisyuk V. S. Application and optimization of the Hough transform for searching for objects in an image. Informatsionnyye sistemy i tekhnologii: Mezhdunarodnyy kongress po informatike [Information Systems and Technologies: Proc. International Congress on Informatics]. Minsk, 2011, part 2, pp. 162–165 (In Russian).
  13. Doermann D. Document Analysis Systems: Theory and Practice. New York, World Scientific Publishing Co. Inc. Publ., 2016, 400 p.
  14. Doermann D. Handbook of Document Image Processing and Recognition. New York, Springer Publ., 2014, 666 p.
  15. Sonka M., Hlavac V., & Boyle R. Image processing, analysis, and machine vision. New York, Cengage Learning Publ., 2014, 520 p.

24.05.2024