Spark testing to measure carbon content in carbon steels based on fractal box counting

Abstract

Spark testing is an effective and inexpensive way to classify the steel materials in industrial practice. Especially, it could determine carbon content in a moment. However, this method has reliance upon the skill and experience of operator. Therefore, computer recognition of spark has been developed to determine carbon content for increasing reliability and efficiency. The numerable high-resolution images of sparks emitting at high speed are taken pictures by camera or mobile phone, then input the images into computer, and calculated the fractal dimension of spark based on fractal box counting method. Result shows that the fractal dimension has a good correlation with carbon content, and the similar natural logarithm function is obtained by a least squares fit through plots of carbon content against fractal dimension. Carbon content measurement error is about 0.06%. It means a convenient and accurate way to identify an unknown carbon steel may be achieved.