DOI QR코드

DOI QR Code

Recognition of Flat Type Signboard using Deep Learning

딥러닝을 이용한 판류형 간판의 인식

  • Kwon, Sang Il (Department of Spatial Information Engineering, Namseoul University) ;
  • Kim, Eui Myoung (Department of Spatial Information Engineering, Namseoul University)
  • Received : 2019.07.22
  • Accepted : 2019.08.24
  • Published : 2019.08.31

Abstract

The specifications of signboards are set for each type of signboards, but the shape and size of the signboard actually installed are not uniform. In addition, because the colors of the signboard are not defined, so various colors are applied to the signboard. Methods for recognizing signboards can be thought of as similar methods of recognizing road signs and license plates, but due to the nature of the signboards, there are limitations in that the signboards can not be recognized in a way similar to road signs and license plates. In this study, we proposed a methodology for recognizing plate-type signboards, which are the main targets of illegal and old signboards, and automatically extracting areas of signboards, using the deep learning-based Faster R-CNN algorithm. The process of recognizing flat type signboards through signboard images captured by using smartphone cameras is divided into two sequences. First, the type of signboard was recognized using deep learning to recognize flat type signboards in various types of signboard images, and the result showed an accuracy of about 71%. Next, when the boundary recognition algorithm for the signboards was applied to recognize the boundary area of the flat type signboard, the boundary of flat type signboard was recognized with an accuracy of 85%.

간판은 유형마다 간판의 규격이 정해져 있으나 실제 설치된 간판은 형태와 크기가 일정하지 않다. 또한, 간판은 간판 내부의 색상에 대한 규정이 정해져 있지 않기 때문에 다양한 색상을 갖고 있다. 간판을 인식하기 위한 방법은 도로표지판과 차량번호판을 인식하는 유사한 방법으로 생각할 수 있으나 간판의 특성으로 인해 도로표지판과 차량번호판과 유사한 방법으로 간판을 인식할 수 없는 한계점이 있다. 이에 본 연구에서는 딥러닝 기반의 Faster R-CNN 알고리즘을 이용하여 불법 및 노후 간판의 주요 대상이 되는 판류형 간판을 인식하고 간판의 영역을 자동으로 추출하는 방법론을 제안하였다. 스마트폰 카메라를 이용하여 촬영한 간판 영상을 통해 판류형 간판을 인식하는 과정은 2가지의 순서로 나뉜다. 먼저, 다양한 유형의 간판 영상에서 판류형 간판을 인식하기 위해 딥러닝을 이용하여 간판의 유형을 인식하였으며 그 결과는 약 71%의 정확도로 나타났다. 다음으로 판류형 간판의 경계영역을 인식하기 위해 간판 영역 인식 알고리즘을 적용하였을 때 85%의 정확도로 판류형 간판의 경계영역을 인식하였다.

Keywords

References

  1. Ahn, H.Y. and Lee, J.T. (2018), Classification of vehicles based on faster R-CNN suitable for use in actual road environments, Journal of Korean Institute of Intelligent Systems, Vol. 28, No. 3, pp. 210-218. (in Korean with English abstract) https://doi.org/10.5391/JKIIS.2018.28.3.210
  2. Cha, S.H. and Kim, J.H. (2017), Analysis on managemental characteristics and problems of outdoor advertisement in Seoul metropolitan city, The Korean Society Of Design Culture, Vol. 23, No. 4, pp. 767-779. (in Korean with English abstract)
  3. Choi, H.S. and Kim, E.M. (2015), Detection of road signs region and recognition of directional information, Proceedings of Korean Society for Geospatial Information Science, September 2015, Korea, pp. 197-198.
  4. Claudio, R.J. and Rodrigo, S. (2004), Rectangle detection based on a windowed hough transform, Proceedings of the XVII Brasilian Symposium on Computer Graphics and Image Processing, October 2004, Brazil, pp. 113-120.
  5. Hua, S. and Xiaoou, T. (2003), Generic sign board detection in images, MIR '03 Proceedings of the 5th ACM SIGMM international workshop on Multimedia information retrieval, 7 July 2003, California, USA, pp. 144-149.
  6. Jang, J.W. and Park, G.M. (2017), License plate recognition system based on normal CCTV, Journal of the Institute of Electronics and Information Engineers, Vol. 54, No. 8, pp. 89-96. (in Korean with English abstract) https://doi.org/10.5573/ieie.2017.54.9.89
  7. Kim, B.J., Kim, D.H., and Lee, J.H. (2016), An improved license plate recognition technique in outdoor image, Journal of Korean institute of intelligent systems, Vol. 26, No. 5, pp. 423-431. (in Korean with English abstract) https://doi.org/10.5391/JKIIS.2016.26.5.423
  8. Ministry of the Interior and Safety(MOIS), (2016), Outdoor Advertising Policy Focus, ISSN 2288-2456, Korea out of home advertising center, Seoul, pp. 1-112.
  9. Ministry of the Interior and Safety(MOIS), (2018), 2018 Outdoor Advertising Statistics, ISSN 2635-5086, Korea out of home advertising center, Seoul, pp. 1-196.
  10. Kwon, S.I. and Kim, E.M. (2019), Recognition of signboard type using faster R-CNN, 2019 KAGIS Spring Conference and Inter-Korean Exchanges and GIS Symposium, 16 May 2019, Chuncheon, Korea, pp. 109-110.
  11. Kwon, S.I. and Kim, E.M. (2019), Recognition of horizontalflat type signboard using Images, 2019 Spring Conference, 31 May 2019, Busan, Korea, pp. 267-268.
  12. Lee, M.S. and Choi, G.S. (2018), Comparative analysis on characteristics of color uses for signboard between before and after signboard improvement project - focused on signboard improvement project in Ganghwa-gun, Incheon, Journal of Korea Society of Color Studies, Vol. 32, No. 2, pp. 51-63. (in Korean with English abstract) https://doi.org/10.17289/jkscs.32.2.201805.51
  13. Lim, H.C., Kaushik, D., and Jo, K.H. (2009), Geometrical reorientation of distorted road sign using projection transformation for road sign recognition, Journal of Institute of Control, Robotics and Systems, Vol. 15, No. 11, pp. 1088-1095. (in Korean with English abstract) https://doi.org/10.5302/J.ICROS.2009.15.11.1088
  14. Tao, Z., Jie, Z., and Wenjing, J. (2018), Fast and robust road sign detection in driver assistance systems, Applied Intelligence, Vol. 48, No. 11, pp. 4113-4127. https://doi.org/10.1007/s10489-018-1199-x