Journal of Sensor Science and Technology (센서학회지)

The Korean Sensors Society (한국센서학회)
권호 표지
DOI QR코드

DOI QR Code

Camera Calibration Using Neural Network with a Small Amount of Data

소수 데이터의 신경망 학습에 의한 카메라 보정

  • Do, Yongtae (School of Electronic & Electrical Engineering, Daegu University)
  • 도용태 (대구대학교 전자전기공학부)
  • Received : 2019.04.30
  • Accepted : 2019.05.28
  • Published : 2019.05.31
Download PDF
⟨ Previous Next ⟩

Abstract

When a camera is employed for 3D sensing, accurate camera calibration is vital as it is a prerequisite for the subsequent steps of the sensing process. Camera calibration is usually performed by complex mathematical modeling and geometric analysis. On the other contrary, data learning using an artificial neural network can establish a transformation relation between the 3D space and the 2D camera image without explicit camera modeling. However, a neural network requires a large amount of accurate data for its learning. A significantly large amount of time and work using a precise system setup is needed to collect extensive data accurately in practice. In this study, we propose a two-step neural calibration method that is effective when only a small amount of learning data is available. In the first step, the camera projection transformation matrix is determined using the limited available data. In the second step, the transformation matrix is used for generating a large amount of synthetic data, and the neural network is trained using the generated data. Results of simulation study have shown that the proposed method as valid and effective.

Keywords

HSSHBT_2019_v28n3_182_f0002.png 이미지

Fig. 2. ANN for learning the projection of a camera.

HSSHBT_2019_v28n3_182_f0003.png 이미지

Fig. 3. Proposed neural camera calibration method.

HSSHBT_2019_v28n3_182_f0004.png 이미지

Fig. 4. Calibration points are marked by ‘+’. They are distorted mainly by the lens and the degree of distortion is represented by the circle size.

HSSHBT_2019_v28n3_182_f0005.png 이미지

Fig. 5. Error circles by the proposed method for the calibration data.

HSSHBT_2019_v28n3_182_f0006.png 이미지

Fig. 1. Pinhole camera model.

Table 1. Camera parameters assumed for test.

HSSHBT_2019_v28n3_182_t0001.png 이미지

Table 2. Lens distortion and noise parameters.

HSSHBT_2019_v28n3_182_t0002.png 이미지

Table 3. Performance(projection error in [mm]) for different number of hidden nodes

HSSHBT_2019_v28n3_182_t0003.png 이미지

Table 4. Comparative test results [mm]

HSSHBT_2019_v28n3_182_t0006.png 이미지

Table 5. Test results using more learning data: Twenty data are used here for calibration while ten data are used in the test of Table 4.

HSSHBT_2019_v28n3_182_t0007.png 이미지

References

  1. T. Jadhav, K. Singh, and A. Abhyankar, "Volumetric estimation using 3D reconstruction method for grading of fruits", Multimed. Tools Appl. Vol. 78, No. 2, pp. 1613-1634, 2019. https://doi.org/10.1007/s11042-018-6271-3
  2. R. Tsai, "A versatile camera calibration technique for highaccuracy 3D machine vision metrology using off-the-shelf TV cameras and lenses", IEEE J. Robot. Autom., Vol. 3, No. 4, pp. 323-344, 1987. https://doi.org/10.1109/JRA.1987.1087109
  3. L. Song, W. Wu, J. Guo, and X. Li, "Survey on camera calibration technique", Proc. IEEE Int. Conf. Intell. Hum. Mach. Syst. Cybern., Hangzhou, China, 2013.
  4. Q. Wang, L. Fu, and Z. Liu, "Review on camera calibration", Proc. Chin. Control Decis. Conf., pp. 3354-3358, Xuzhou, China, 2010.
  5. S. Walczak, "Artificial neural networks," in Advanced Methodologies and Technologies in Artificial Intelligence, Computer Simulation, and Human-Computer Interaction, M. Khosrow-Pour, Ed., IGI Global, Hershey, pp. 40-53 2018.
  6. D.-M. Woo and D.-C. Park, "An efficient method for camera calibration using multilayer perceptron type neural network", Proc. Int. Conf. Future Comput. Commun., pp. 358-362, Kuala Lumpar, Malaysia, 2009.
  7. J. Xiong, J. Xia, X. Xu, and Z. Tian, "A novel method of stereo camera calibration using BP neural network", Appl. Mech. Mater., Vol. 29-32, pp. 2692-2697, 2010. https://doi.org/10.4028/www.scientific.net/AMM.29-32.2692
  8. Q. Memon and S. Khan, "Camera calibration and threedimensional world reconstruction of stereo-vision using neural networks", Int. J. Syst. Sci., Vol. 32, No. 9, pp. 1155-1159, 2001. https://doi.org/10.1080/00207720010024276
  9. B. Chen, W. Wang, and Q. Qin, "Stereo vision calibration based on GMDH neural network", Appl. Opt., Vol. 51, No. 7, pp. 841-845, 2012. https://doi.org/10.1364/AO.51.000841
  10. X. Chen, H. Fang, Y. Yang, and S. Qin, "The research of camera distortion correction based on neural network", Proc. Chin. Control Decis. Conf., pp. 596-601, Mianyang, China , 2011.
  11. Y. Do, J. Neubert, N. Ferrier, and Y. Hu., "Camera calibration where practical uncertainties exist in camera model and calibration data", Proc. IEEE Conf. Mechatron. Mach. Vis. Pract., pp. 491-495, Hong Kong, 2001.
  12. Y. Do, "Neural camera calibration using a limited number of data", Proc. Korean Sens. Soc. Conf. Vol. 29, p. 217, Seoul, Korea, 2018.
  13. M.T. Hagan and M. Menhaj, "Training feedforward networks with the Marquardt algorithm", IEEE Trans. Neural Netw., Vol. 5, No. 6, pp. 989-993, 1994. https://doi.org/10.1109/72.329697