Qualicision® - Relation

CVB-based analysis module for calculating Fuzzy-Similarities between two criterion vectors

Similarity evaluation of image criteria based on signal curves, which may represent gray scale value, length, angle, discontinuity of gray scale value, etc.

Similarity: 0.95

The CVB-based analysis module Qualicision® - Relation allows the comparison of any two signal curves, the result of which indicates a similarity degree with values belonging to the interval [0,1], where 1 indicates identical signal curves and any value <1 a gradual reduction of identity.

The characteristic signal curve of any image may represent one criterion, as for example a gray scale value histogram, or composed criteria, as for example the number of pixels in the intervals [0,50], [51,70], [71,100], [101,150], [151,255] and their mean gray value. When creating characteristic signal curves, it is possible to reduce or enhance the importance of single criteria, which allows dealing separately with existing or lacking image criteria.

Technical data:

• Compatible with the CVB-Concept
• Asymmetric, positive and negative similarity relations
• Optional: evaluation of any vector length with components from the [0,1] interval
• Running under Windows NT

Reference application: Hengst Filterwerke GmbH & Co.KG

The flexible and free configurable software FuzzyImageDesk® is installed at the end of the production line at automotive supplier Hengst Filterwerke GmbH & Co.KG, Münster. In time intervals of less than 4 seconds, FuzzyImageDesk® checks a range of about 15 different filter units for completeness and correct fixture of their respective components. The visualized control results are recorded and then transmitted to the SPC.

During the production process, each filter unit is mounted on a rack and conveyed from one assembly or control station to the next. Automation on one side and increased quality requirements from the automotive industry on the other forced the automotive provider to replace human quality controllers by conventional control methods as light barriers, pressure testers or probe tips. However, given the complex filter structures, it was still not possible to grant a 100% control.

Implementing Fuzzy Image-Analysis, the desired automated overall control became possible, analyzing both filter unit surface and completeness. At Hengst Filterwerke, Fuzzy Image-Analysis is part of an additional control station at the end of the production line, checking completeness and correct fixture of the assembled components. As the production line cycle time is < 4 seconds, within this time a "good-bad-decision" has to be transmitted to the SPC, so any bad unit can be sorted out.