Advertisement

FPGA Implementation of a Video Based Abnormal Action Detection System with Real-Time Cubic Higher Order Local Auto-Correlation Analysis

  • Kaoru Hamasaki
  • Keisuke Dohi
  • Yuichiro Shibata
  • Kiyoshi Oguri
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 8405)

Abstract

In this paper, we show FPGA implementation of a real-time video-based abnormal action detection system, which is a key basic function of applications such as security systems and monitoring systems for nursing elderly people. Our system extracts Cubic Higher order Local Auto-Correlation (CHLAC) features from input video frames and detects abnormal actions with a subspace method based on Candid Covariance-free Incremental Principal Component Analysis (CCIPCA). Empirical experiments demonstrate our system works well at 62.5 fps, which is limited by a camera device. The system implemented on the FPGA is estimated to achieve up to 240 fps, which corresponds to 8.6 times speedup compare to software execution on a PC. It is also shown that the FPGA implementation is more than 20 times energy efficient than the software execution.

Keywords

Subspace Method FPGA Implementation Software Execution Machine Vision Application Pipeline Schedule 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Otsu, N.: Towards flexible and intelligent vision systems-from thresholding to CHLAC. In: IAPR Conference on Machine Vision Application, pp. 430–439 (2005)Google Scholar
  2. 2.
    Nanri, T., Otsu, N.: Unsupervised abnormality detection in video surveillance. In: IAPR Conference on Machine Vision Applications, pp. 574–577 (2005)Google Scholar
  3. 3.
    Weng, J., Zhang, Y., Hwang, W.S.: Candid covariance-free incremental principal component analysis. IEEE Trans. on Pattern Analysis and Machine Intelligence 25(8), 1034–1040 (2003)CrossRefGoogle Scholar
  4. 4.
    Ishii, I., Sukenobe, R., Yamamoto, K., Takaki, T.: Real-time image recognition using hlac features at 1000 fps. In: Proceedings of the 2009 International Conference on Robotics and Biomimetics, pp. 954–959 (2009)Google Scholar
  5. 5.
    Shiraki, T., Saito, H., Kamoshida, Y., Ishiguro, K., Fukano, R., Shirai, T., Taura, K., Otake, M., Sato, T., Otsu, N.: Real-time motion recognition using chlac features and cluster computing. In: Proceedings of the 3rd IFIP International Conference on Network and Parallel Computing, pp. 50–56 (2006)Google Scholar
  6. 6.
    Dohi, K., Yorita, Y., Shibata, Y., Oguri, K.: Pattern compression of FAST corner detection for efficient hardware implementation. In: Proc. IEEE 21st Int. Conf. Field Programmable Logic and Applications, pp. 478–481 (September 2011)Google Scholar
  7. 7.
    Negi, K., Dohi, K., Shibata, Y., Oguri, K.: Deep pipelined one-chip FPGA implementation of a real-time image-based human detection algorithm. In: Proc. Int. Conf. Field-Programmable Technology, pp. 1–8 (December 2011)Google Scholar
  8. 8.
    Dohi, K., Hatanaka, Y., Negi, K., Shibata, Y., Oguri, K.: Deep-pipelined FPGA implementation of ellipse estimation for eye tracking. In: Proc. IEEE 22st Int. Conf. Field Programmable Logic and Applications, pp. 458–463 (2012)Google Scholar

Copyright information

© Springer International Publishing Switzerland 2014

Authors and Affiliations

  • Kaoru Hamasaki
    • 1
  • Keisuke Dohi
    • 1
  • Yuichiro Shibata
    • 1
  • Kiyoshi Oguri
    • 1
  1. 1.Nagasaki UniversityNagasakiJapan

Personalised recommendations