The Cucumber Cognizance for Back Propagation of Nerual Network

신경회로망의 오류역전파 알고리즘을 이용한 오이 인식

  • Min, Byeong-Ro (Dept. of Bio-Mechatronic Engineering, Sungkyunkwan National University) ;
  • Lee, Dae-Weon (Dept. of Bio-Mechatronic Engineering, Sungkyunkwan National University)
  • 민병로 (성균관대학교 바이오메카트로닉스학과) ;
  • 이대원 (성균관대학교 바이오메카트로닉스학과)
  • Received : 2011.08.22
  • Accepted : 2011.11.18
  • Published : 2011.12.31

Abstract

We carried out shape recognition. We found out cucumber's feature shape by means of neural network and back propagation algorithm. We developed an algorithm which finds object position and shape in real image and we gained following conclusion as a result. It was processed for feature shape extraction of cucumber to detect automatic. The output pattern rates of the miss-detected objects was 0.1~4.2% in the output pattern which was recognized as cucumber. We were gained output pattern according to image resolution $445{\times}363$, $501{\times}391$, $450{\times}271$, $297{\times}421$. It was appeared that no change was detected. When learning pattern was increased to 25, miss-detection ratio was 16.02%, and when learning pattern had 2 pattern, it didn't detect 8 cucumber in 40 images.

정확한 오이의 형상 및 위치를 인식하기 위하여 형상인식 알고리즘에 대한 연구를 수행하였다. 실제 영상에서 오이의 형상과 위치를 판정할 수 있도록 알고리즘을 개발한 결과, 다음과 같은 결론을 얻었다. 오이의 특징형상 검출은 $15{\times}15$ 간격으로 자동검출 되도록 처리하였다. 오이로 인식된 출력패턴 중에서 오검출된 출력패턴의 비율은 0.1~4.2%로 나타났다. 오류역전파 알고리즘은 영상크기를 $445{\times}363$, $501{\times}391$, $300{\times}421$, $450{\times}271$, $297{\times}421$의 크기에 따라 출력패턴을 얻은 결과 영상의 크기에 따른 검출 값의 변화는 없는 것으로 나타났다. 학습패턴 수가 25개로 증가하면 영상에서 다른 패턴을 검출하는 비율이 16.02%로 나타났다. 또한 학습패턴이 2개인 경우 40개의 영상에서 8개의 오이를 검출하지 못하였다. 학습패턴의 수가 7~9개인 경우 오이의 검출이 가장 좋은 것으로 나타났다.

Keywords

References

  1. Kim, S.C., D.Y. Choi, and H. Hwang. 2001. Automation of bio-industrial process via tele-task command (I). Journal of the Korean Society for Agricultural Machinery 26(1):21-28 (in Korean).
  2. Kim, Y.S., B.R. Min, W. Kim, D.W. Kim, K.W. Seo, H.T. Kim, and D.W. Lee. 2002. Development of image processing system for tracking location of moving objects. Proceedings of the KSAM 2002 7(2):292-297 (in Korean).
  3. Kondo, N., Y. Nishitsuji, P.P. Ling, and K.C. Ting. 1996. Visual feedback guded robotic cherry tomato harvesting. Transactions of the ASAE 39(6):2331-2338.
  4. Lee, D.W., B.R. Min, H.T. Kim, K.T. Lim, W. Kim, Y.S. Keon, Y.L. Nam, J.W. Chon, and S.H. Sung. 1998. Effects of cultivation method on the growth and yield of a cucumber for development of a robotic harvester. Journal of Bio-Environment Control 7(3):235-245 (in Korean).
  5. Lee, Y.K., D.W. Lee, D.W. Kim, H.T. Kim, B.R. Min, and W. Kim. 2001. Development of image processing system for a harvesting robot of a grape. Proceedings of the KSAM 2001 6(1):181-186 (in Korean).
  6. Masaru Hifa, Seiji Nakao, and Toshio Iwao. 1995. Vision sensor for cherry tomato harvesting robot. JSAM Vol. 1, November 13-20.
  7. Min, B.R. and D.W. Lee. 2004. Development of the robot manipulator for kinematics. Journal of Bio-Environment Control 13(1):1-7 (in Korean).
  8. Min, B.R., J.H. Mun, and D.W. Lee. 2003. Development of an end-effector for cucumber robotic harvester. Journal of Bio-Environment Control 12(2):63-67 (in Korean).
  9. Min, B.R., J.H. Mun, and D.W. Lee. 2003. Development of the manipulator of a cucumber robotic harvester. Journal of Bio-Environment Control 12(2):57-62 (in Korean).
  10. Min, B.R., K.T. Lim, and D.W. Lee. 2011. The research of shape recognition algorithm for image processing of cucumber harvest robot. Journal of Bio-Environment Control 20(2):1-9 (in Korean).
  11. Seo, K.W., B.R. Min, W. Kim, D.W. Kim, C.W. Lee and D.W. Lee. 2003. The dynamic stereo vision system for recognizing a position in the 3-D coordinates. Proceedings of the KSAM 2003 8(1):478-483 (in Korean).
  12. Shiller, Zvi and Steven Dubowsky. 1989. Robot path planning with obstacles, actuator, gripper, and payload constraints, The Intermational Journal of Robotic Research 8(6):3-18. https://doi.org/10.1177/027836498900800601