Automated image inspection system to quantify thermal spray splat morphology

Abstract

Nowadays, coating materials such as ceramic, metal or alloys, are widely used in the industry to improve properties of the base materials. There is growing interest in knowing the parameters governing the promotion of high quality coatings, mainly adhesion strength. It would be really interesting using Automated Measurement Tools based on Image Processing procedures in order to obtain an image inspection system, owing to the fact that these techniques are contrasted tools in other inspection and quality control systems. In this work alumina splats were plasma-sprayed onto an aluminium surface under atmospheric conditions. Top view images of these splats were obtained by a scanning electron microscope and processed by Image Processing Techniques. As the microstructure of the splats forming a coating is a very important factor which determines the adhesion strength of the coated piece, this system was developed to characterise these splats in terms of their area, perimeter, circularity, eccentricity and solidity. As a result, our system can be used as the initial stage of an automated inspection system, which could be helpful to a human expert, to take a decision whether the spraying splats, and hence the spraying parameters, are the appropriated ones. In a later stage, our interest will be to substitute this human expert for an expert system based on the knowledge theory, which would be able to decide about the quality of these splats without human intervention. In this work, an automated image inspection system to assist human experts to quantify thermal spray splat morphology is shown. The automatic system was able to classify the coating by means of a decision tree, foretelling the suitability of each type of splats in a coating, and thus the spraying process parameters utilised.