On the application of digital image processing to surfaces of different nitride coatings

In this work the application of different image processing techniques to low thickness coatings of CrN, NbN and TaN produced by unbalanced magnetron sputtering has been employed, in order to get an alternative methodology for calculating characteristics such as roughness, grain size and number of grains. The roughness can be compared with the degree of complexity or tortuosity and it can be indicated using the fractal dimension. Grain size is normally determined by measuring the diameter of many grains and averaging these values. Once the features had been calculated, a clustering process was implemented for classifying images of different materials employing the characteristics mentioned before. For this classification, the K-Means method was employed, and its performance was evaluated for each material (class), grouping the characteristics in five different ways. The conclusion was that CrN coatings were the best classified, contrasting with NbN coatings, which were difficult to classify.