Visual Inspection Systems

Abstract

Following the previous chapter, the first part of this chapter again deals with quality control and fault detection aspects. However, instead of supervising raw process data (measurement signals), here we deal with two-dimensional visual context information in form of images. In particular, in Section 9.1 we demonstrate how images showing the surface of production items will be analyzed if they show any untypical occurrences which may indicate faults on single items, usually caused by failures in the production systems. The second part of this chapter deals with visual context information in form of textures, which can be seen as special types of images containing repeating patterns over their complete range. Human perception modelling for visual textures will be the core aspect of the second part. This means that models will be built for the purpose to associate human perception and emotions with given visual textures. These association models require a visual texture as input and are able to respond to a certain emotion the input texture may trigger for customers.