Filter Orientation

A Machine Vision System for Pose Estimation and Part Verification Based on Eigenimages

A PC-based machine vision system is designed for precise rotational positioning and reliable verification of industrial parts. The system has been integrated into the production line, which is capable of assembling various types of automobile oil filters, each one having its own visual appearance. Our primary goal was to design a flexible system, which could be easily adapted for assembling several different filter types. To achieve this an appearance based approach, employing the Karhunen-Loeve expansion, was used. With this method, the most significant visual information is automatically extracted from a set of rotated filter images, i.e. templates, and described by a small number of eigenimages, which constitute the eigenspace. The templates are then projected to the eigenspace. Given a captured image of the filter in an unknown position the system projects the image into the eigenspace and the distances between the projected templates and the projected filter image are computed. Based on the minimum distance the filter position is determined, its type is verified, and the filter is rotated. The system operates in a closed loop, therefore the new position can be evaluated and corrected, if required. The results obtained show that the system works reliably, and meets the required accuracy and speed.

Publication:

Franci Lahajnar, Stanislav Kovačič, "Machine vision system for positioning and part verification of gas oil filters based in eigenimages". V: Tobin, Kenneth W. (ed.). Machine vision applications in industrial inspection VIII : 24-26 January 2000, San Jose, California, (SPIE proceedings series, vol. 3966). Bellingham (Washington): SPIE-The International Society for Optical Engineeering, cop. 2000, str. 220-227.