Home | english  | Impressum | Sitemap | KIT

Automatische Sichtprüfung gehonter Oberflächen

Automatische Sichtprüfung gehonter Oberflächen
Ansprechpartner:

Limeng Wang

Projektgruppe:

Bildverarbeitung

Introduction
Currently image based optical measurement is proved to be an efficient und safe technology for the quality control round about automotive motors. Honed surface in cylinder bores, as one of the measured objects, plays an essential role with respect to oil consumption, noxious emissions, and running performance. The characterization of honed surface relates to line structured texture. How to access such complex technical surface is until now still an interesting topic for the researcher. The approaches, which aim to extract the measures of honing texture and inspect surface defects, have been studied in a long-term and especially achieved success for an offline-inspection after entire honing process is completed. Unfortunately a solution to automate surface manufacturing with online-inspection is nowadays not available in industrial environment. Our research work intends to develop a novel image analysis method to monitor the state of honed surfaces and feed back the results to the honing machine, as shown in Fig. 1.
                      
General problems
In this project 2D microscopic/macroscopic image acquisition with CCD sensor works as data foundation for image processing. Some challenges for texture recognition are demonstrated by both examples of incompletely and completely honed surface in Fig. 2. Cracks in any direction and confused edges make it difficult to distinguish honing grooves from defects. Correspondently the direction of honing grooves is interrupt by flakes and material smearing, so that the position of single groove becomes unclear. Although a few parts of honing texture by human vision is perhaps still identifiable,  existing model-based algorithms are unable to extract effective features of the line structure in computer. Out of this motivation some additional possibilities to characterizing honing grooves should be taken into account.

           

                                           Fig. 1:  Structure of an automated honing process with the vision system

 

                                      

                                      Fig. 2: (1) left: incompletely honed surface (2) right: completely honed surface

 

Project objective

In order to raise the performance for surface inspection in this project, our research refers to following steps:

  • Image enhancement for an evident display of image structure
  • Effective and robust extraction of features, which supplement each other to describe honing texture
  • Positioning honing grooves
  • Segmentation of image signals into defect and groove structures
  • Inspection of defects, such as groove interrupts, folded metal etc. and measurement of groove properties, such as the crosshatch angle and groove densities
  • Overall assessment for surface quality
  • Automating the vision system with integrated algorithm library

Implementation for robust and accurate measurement on honed surface is finally desired under the circumstances of different surface qualities.