Figure 2.Image acquisition system.The system provides adaptive but rigid joints between its components which, together with the control of the position of the table and the camera, allows us to carry out studies of repeatability in the capture and the subsequent analysis.The image acquisition system is based on the phenomenon of light retroreflection [5,6], that enhances the existing distortions on a flat surface. This kind of method has been applied to enhance the surface variations in different fields, such as plastic materials [7,8], aerospace industry [9,10], squeezing processes in general [11], and the automobile industry in particular [5,12].3.?General Description of the Algorithms for the Determination of the Geometrical Parameters of the ImperfectionsTo characterize the imperfection, an algorithm has been developed that extracts the information from the imperfection once it is firstly selected.

There are algorithms that try the surface reconstruction using least-squares techniques [13], but they do not allow the numerical quantification of imperfections of a size around 20 mm, such as the ones to be analyzed in this case. In [14] a collection of algorithms for the analysis of textures is shown, but most of them are for the classification of patterns. The work in [15] deals with a method based on the evaluation of local wall thickness and other imperfections, such us creases, using a medial axis transformation.All the information regarding the sheet deformation is available in the profile obtained from the captured image.

Taking into account that this information is the variation of the sheet deformation, the profile (signal) to GSK-3 work on is the derivative of the original profile. The inflection points of this signal coincide with the maxima and minima of the original profile (sheet profile). The light areas of the image coincide with the prominences and the dark ones with the valleys of the sheet. The information of the highest elevation of the imperfection is related to the value of the maximum gradient of the signal. To track the imperfection, the maximum deflection is followed. Therefore, the algorithm must search for this maximum gradient and move along the imperfection with this searching criterion.The algorithms developed for the determination of the geometrical parameters of the imperfection and its quality index are:Algorithm for the obtaining of the profiles.

Algorithm for the automated monitoring of the imperfection.Extraction of the parameters of the imperfection and classification.4.?Algorithm for the Obtaining of Profiles4.1. Profile AnalysisThe first step to take is the analysis of the profile obtained from an image. The profile is typically of the form shown in Figure 3b. To get this profile, firstly a media filter is applied in the direction in which the imperfection is analyzed.