Sink Mark Detection on Highly Reflective Plastics Surfaces

Research output: ThesisDoctoral Thesis


The appearance of plastics products has become one of the major factors for the buying decision of customers. Especially curved high-quality front panels with mostly black high-gloss surface enhance the look of products. However, sink marks are prominent on such surfaces, even when their depth is in the lower range of microns. In the given work, sink mark detection which is based on phase measuring deflectometry is evaluated in a laboratory testing site. Additionally, a concept design for the integration of such a testing site into an injection molding process line is introduced. In this design, a unit for sample positioning was added to the basic deflectometric setup to enable total inspection of curved high-gloss surfaces. For evaluation of the inspected surface, a model was developed to calculate sink mark intensity from sink mark curvature and area. Therein, polynomial base functions were implemented in the image processing algorithms. The main application of the polynomials was the smoothing and differentiation of the data. This approach allowed for the robust calculation of surface curvature from the deflectometric results which are proportional to the first derivative of the surface. The results for sink mark intensity were compared to topographic measurements which were obtained by a chromatic coding confocal microscope. The topographic measurements were modelled with a Gaussian-like function. The function allows for description of sink mark shape with few parameters and facilitates the calculation of features which can be used to evaluate sink mark visibility. Despite the fact that the two measurement methods are based on different principles, the results of topographic and deflectometric measurements show high correlation. For this reason, deflectometric measurements of sink marks on at least black high-gloss surfaces can be calibrated with data from topographic measurements.


Translated title of the contributionEinfallstellendetektion auf hochreflektierenden Kunststoffoberflächen
Original languageEnglish
Publication statusPublished - 2014