SLP Header

Faulty Gear Identification in Automobile industries using Watershed Segmentation

IJCSEC Front Page

In previous years industries faced a major problem in testing their final finished products especially in the field of automobiles; it reduces the efficiency of the vehicles when the parts are not tested efficiently. Yet the selection and rejection of this automobile part for the further implementation in automobiles is needed. To overcome this, industries started using mechanical instruments to test these automobile parts they are known as go, no-go instruments. Each object has a separate instruments as per its measurements specifications if it get fitted in the respective instrument exactly then that finished product has no defects. If that object doesn’t get fitted in the respective instruments then that finished product is said to have the defect and it has to be rejected. The disadvantage faced by this technique is that the measuring instruments gets wears, so as using this wearer instrument for further testing its accuracy is lost. Elimination of this 0.1% of error of the small automobile parts can cause major accidents in case of large heavy vehicles and more over the life of the vehicle is reduced.
Worldwide, around 1.2 million people died as a result of accidents in 2012. This represents an average of 3,242 persons dying each day around the world. In addition to these deaths, around 50 million people globally are estimated to be injured or disabled every year. Projections indicate that these figures will increase by about 65% over the next 20 years. Road accidents are currently world's eleventh leading cause of death, but by 2020, it will become third, behind deaths linked to heart disease and mental illness. In United States alone, around 6.2 million traffic accidents occur due to automobile crashes in 2012. And about 1million accidents occur due to the poor working of automobile parts. These accidents accounted for 42,636 deaths and 2,788,000 nonfatal injuries. Thus in some countries image processing technique came into the existence for testing the faults in the automobile parts. This uses a camera to capture the image of the products. Standard image of the product is fed to the pc. By using suitable programming techniques the standard image is compared with the real time image. If the real time image doesn’t matches the standard image then it is said to have a fault. However this technique too have a by using suitable programming techniques the standard image is compared with the real time image. If the real time image doesn’t matches the standard image then it is said to have a fault.
However this technique too have a disadvantage that it could find only the 2D faults and individual device set up is required for each objects. This project overcomes all the above disadvantages and hence it is a 3D fault detection for multiple products in a single device set up using upcoming technology, artificial neural network. Various researches are being carried out for the proper identification selection, and rejection of damaged parts.
To overcome the disadvantages of the testing process in automobile industries image processing is being used to test the faults in the objects. The project “Test automation in automobile parts using image processing” is used to overcome the difficulties faced in the field of testing and fault detection in automobile industries. The major problem in the existing system is that it can be used only for fault detection in two dimensional image processing. Another major disadvantage is that separate device set up is needed for individual objects which increases the cost of production and reduces accuracy.


  1. Surface Inspection of polyurethane foam by image processing and Artificial Neural Networks Techniques ArantzaBereciartua, josé angel gutiérrezolabarria, InfoTech dept. Robotiker, corporations’ technologic tecnalia. parquet technological, edificio 202.e-48170 zamudio (bizkaia)
  2. Method of evaluating object based upon image processing and inspection apparatus using said method US 5638460 A
  3. “Fault Identification and Fault-Tolerant ControlforNonlinearSystemsBasedon Fuzzy T-S Model “,Wei Wang ; Shaocheng Tong ; Yang YuInnovative Computing, Information and Control (ICICIC), 2009 Fourth International Conference onDigital ObjectIdentifier:10.1109/ICICIC.2009.205 Publication Year: 2009 , Page(s): 946 - 949
  4. E. J. Candès and D. L. Donoho, “New tight frames of curvelets and optimal representations of objects with piecewise C2 singularities”,Comm. on Pure and Appl. Math, Vol. 57, No. 2 , pp. 219–266, 2004.
  5. “A Method of Gear Fault Detection based on Wavelet Transform”, Xiangyu Zhao Business Intelligence and Financial Engineering, 2009. BIFE '09. International Conference onDigital Object Identifier: 10.1109/BIFE.2009.151 Publication Year: 2009, Page(s): 648 - 651
  6. “Fault identification and classification of rolling element bearing based on time-varying autoregressive spectrum”byGuofeng Wang, Mechanical Systems and Signal ProcessingVolume 22, Issue 4, May 2008, Pages 934–947 Special Issue: Crack Effects in Rotordynamics
  7. “Recognition ofboltandnutusingartificialneuralnetwork“,Johan, T.M. ; Prabuwono, A.S.Pattern Analysis and Intelligent Robotics (ICPAIR), 2011 International Conference on Volume: 1 Digital Object Identifier: 10.1109/ICPAIR.2011.5976889 Publication Year: 2011 , Page(s): 165 – 170
  8. Hui Li, Yuping Zhang and HaiqiZheng. “Wear Detection in Gear System Technology, vol.20, pp.1781-1789, November 2006.
  9. LaurentNajman, Michel Schmitt, “Geodesic saliency of watershed contours and hierarchical segmentation”, IEEE Trans. Patt. Anal. Machine Intell. 1996, 18(12):1163-1173.
  10. “An Improved FractalDimension Algorithm and its Application on GearFault Recognition “,Han XiaoFuzzy Systems and Knowledge Discovery, 2009. FSKD '09. Sixth International Conference on Volume: 5, Digital Object Identifier: 10.1109/FSKD.2009.508 Publication Year: 2009.